Programmable C-to-U RNA editing to recruit endogenous APOBEC using engineered RNA

Abstract

The present application relates to a kind of RNA base editing method and editing tool for recruiting endogenous APOBEC family enzyme by APOBEC recruiting guide RNA (APOBEC-recruiting guide RNA, rgRNA) to convert target cytosine into uridine. The rgRNA includes: at least one APOBEC recruiting domain and at least one programmable antisense region, when the rgRNA is combined with the target RNA, an editing loop that induces a circular single-stranded structure on the target RNA sequence. In the way of "catch recruit execute", accurate and flexible base conversion is realized, and the boundary of RNA editing in gene regulation and therapeutic application is expanded.

Description

Technical Field

[0001] This invention relates to the field of molecular biology, specifically to an RNA editing technology and its applications, and particularly to a programmable C-to-U RNA editing tool that uses engineered RNA to recruit endogenous APOBECs and its applications. Background Technology

[0002] RNA editing is widespread in eukaryotes and involves temporary nucleobase alterations after transcription, primarily including A-to-I and C-to-U deamination reactions. This process forms an important regulatory layer, finely modulating various cellular processes under physiological and pathological backgrounds. RNA editing allows for the reversible reprogramming of posttranscribed genetic information, thereby participating in cellular developmental adaptation, cellular stress responses, tissue construction, and the maintenance of homeostasis. Reusing endogenous mechanisms for programmable base editing of cellular RNA holds promise for avoiding the abnormal effector activity, delivery barriers, and immunogenicity caused by the non-physiological expression of exogenous editors, offering a specific and flexible strategy for basic research and disease treatment.

[0003] As our understanding of the mechanisms of natural deaminases deepens, natural editors from the ADAR (adenosine deaminasesacting on RNA) and APOBEC (apolipoprotein B mRNA editing catalytic polypeptide-like) families have been adapted into various efficient site-directed RNA editing platforms. Previous systems typically integrate the catalytically active domains of ADAR or APOBEC with programmable RNA targeting modules, including CRISPR-based systems, non-CRISPR ribonucleoprotein systems, and single-protein systems. By reversibly and dose-tunably rewriting arbitrary adenosine or cytidine in target transcripts, site-directed RNA editing not only complements genome editing in correcting genetic mutations but also provides a powerful tool for regulating signaling pathways, metabolism, and other clinically relevant processes in non-hereditary diseases. However, ectopic expression of exogenous editors poses significant risks, including off-target effects across the genome and transcriptome, immunogenicity, potential oncogenicity, and difficulties in sustained in vivo delivery. These issues limit the translation of related technologies into clinical applications.

[0004] To address the aforementioned issues, a promising new strategy has emerged in recent years: programmable A-to-I editing through engineered guide RNAs that simultaneously bind to target transcripts and recruit endogenous ADARs. This approach eliminates the need for exogenous proteins, deploying as a single RNA molecule. It is simple, specific, and scalable, and its paradigm is similar to clinically validated nucleic acid drug paradigms such as siRNA and ASO, enabling nucleic acid drugs to move further from transcript knockdown to precise base editing. The robustness of this type of ADAR-recruiting RNA editing relies primarily on the hybridization of the guide RNA and target transcript to form a double-stranded RNA structure, which precisely mimics the natural substrate of ADAR. This intricate structural match fundamentally limits the direct extension of this strategy to other RNA-modifying enzymes. For example, APOBEC prefers to catalyze C-to-U conversions on single-stranded RNA substrates, thus the double-stranded RNA-dependent design of ADAR cannot be simply applied. Therefore, to date, efficient single-RNA recruitment editing remains primarily limited to adenosine-to-inosine (A-to-I) base editing. Summary of the Invention

[0005] To address the aforementioned problems in existing technologies, this invention provides an RNA editing method called RNA editing byversatile guide RNA engineering for recruiting specific endogenous APOBECs, abbreviated as REVERSE. This method utilizes APOBEC recruiting guide RNA (rgRNA) to recruit endogenous APOBEC family enzymes. The rgRNA guides endogenous APOBECs, especially APOBEC3A (A3A), to perform precise RNA editing, converting target cytidine to uridine. The rgRNA contains a programmable antisense region complementary to the target RNA (responsible for "capture") and an APOBEC recruitment domain derived from natural APOBEC substrates (responsible for "recruitment"), and can induce the formation of a circular single-stranded structure in the target RNA, serving as a substrate for APOBEC-mediated C to U editing (responsible for "execution"). To expand the editing scope, the inventors further engineered the rgRNA, enabling controlled, multiple cytidine deamination within a well-defined editing loop, and utilizing the difference in substrate preferences between human APOBEC and ADAR to achieve simultaneous editing of cytidine and adenosine. This allows the system to be used both to correct pathogenic mutations and to regulate protein function.

[0006] Specifically, On one hand, the present invention provides an APOBEC recruitment guide RNA (rgRNA), comprising: At least one APOBEC recruitment domain that can bind to the APOBEC protein; and At least one programmable antisense region, each programmable antisense region including an operably linked first antisense region and a second antisense region, wherein the first antisense region and the second antisense region are complementary to the first binding region and the second binding region of the target RNA sequence, respectively; When rgRNA binds to target RNA, a circular single-stranded editing loop is formed between the first and second binding regions on the target RNA sequence. The editing loop has a 5'-N1CN2-3' motif, where N1 and N2 are any ribonucleotides and cytosine C is the base editing site.

[0007] Furthermore, the APOBEC recruitment guide RNA (rgRNA) of the present invention is characterized in that the APOBEC recruitment domain includes a stem-loop secondary structure formed by a single-stranded loop and a palindromic arm, and is selected from the APOBEC recruitment domain of a natural APOBEC substrate, or artificially modified based on the APOBEC recruitment domain of a natural APOBEC substrate.

[0008] Furthermore, the APOBEC recruitment guide RNA (rgRNA) of the present invention is characterized in that the APOBEC recruitment domain is a variant artificially modified based on the APOBEC recruitment domain of the natural APOBEC substrate; The length of the palindromic arm is 2-9 bp, and the single-stranded ring is an NAUC tetranucleotide ring, where N is any ribonucleotide; Preferably, the length of the palindromic arm is 5 bp, and the single-chain loop is selected from CAUC and UAUC; Preferably, the APOBEC recruitment domain has one or more structural units selected from SEQ ID NO:1-5, or has one or more structural units formed by nucleotide deletion, substitution, or addition based on SEQ ID NO:1-5; For example, the APOBEC recruitment structural domain may have 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 or more structural units; Most preferably, the APOBEC recruitment domain is shown in SEQ ID NO:6.

[0009] Furthermore, the APOBEC recruitment guide RNA (rgRNA) of the present invention is characterized in that the length of the programmable antisense region is 20-200 nt, preferably 100-150 nt, more preferably 150 nt; the adenine in the programmable antisense region can be replaced with 2,6-diaminopurine.

[0010] Furthermore, the APOBEC recruitment guide RNA (rgRNA) of the present invention is characterized in that the APOBEC protein is selected from APOBEC1 (A1), APOBEC3A (A3A), and APOBEC3G (A3G). Preferred, When the editing ring has a 5'-ACN2-3' motif, the APOBEC recruitment structure domain combined with APOBEC1 (A1) is used; When the edit ring has a 5'-UCN2-3' motif, the APOBEC recruitment structure domain combined with APOBEC3A (A3A) is used; When the edit ring has a 5'-CCN2-3' motif, the APOBEC recruitment structure domain combined with APOBEC3G (A3G) is used.

[0011] Furthermore, the APOBEC recruitment guide RNA (rgRNA) of the present invention is characterized in that the editing loop is a 4-24 nt circular single strand, preferably a 16 nt circular single strand.

[0012] Furthermore, the APOBEC recruitment guide RNA (rgRNA) of the present invention is characterized by further comprising one or more mRNA binding sequences (RBS), wherein the RBS binds to the region containing editable cytosine outside the first binding region and the second binding region on the target RNA sequence, so as to reduce the off-target editing of the target RNA editing loop cytosine by the APOBEC protein.

[0013] Furthermore, the APOBEC recruitment guide RNA (rgRNA) of the present invention is characterized in that an APOBEC recruitment domain is present between the RBS and the programmable antisense region; and an APOBEC recruitment domain is present between the plurality of RBSs.

[0014] Furthermore, the APOBEC recruitment guide RNA (rgRNA) of the present invention is characterized in that it is a linear rgRNA or a circular rgRNA; preferably a circular rgRNA, wherein the circular rgRNA is a covalently circularized RNA formed by placing flexible RNA linkers on both sides of a linear rgRNA.

[0015] In a second aspect, the present invention provides an RNA editing tool, which includes the APOBEC recruitment guide RNA (rgRNA) described in the first aspect of the present invention, or is capable of generating the APOBEC recruitment guide RNA (rgRNA) described in the first aspect of the present invention. Preferably, the RNA editing tool is selected from plasmids, LNP-rgRNA, and viral vectors; More preferably, the plasmid or viral vector can be transcribed into rgRNA within the cell, and the LNP-rgRNA can deliver rgRNA into the cell.

[0016] Furthermore, the RNA editing tool of the present invention may also include an ADAR recruitment domain, which is capable of recruiting endogenous ADAR and achieving A-to-I editing of target genes; Preferably, the sequence of the ADAR recruitment domain is selected from SEQ ID NO:7-8.

[0017] Thirdly, the present invention provides the application of the APOBEC recruitment guide RNA (rgRNA) described in the first aspect or the RNA editing tool described in the second aspect in correcting pathogenic gene mutations and / or inhibiting the expression of pathogenic genes, wherein the application does not alter the transcript level of the target gene.

[0018] Furthermore, the application described in this invention is characterized by generating PTC by introducing C-to-U into the mRNA sequence of the pathogenic gene, thereby reducing the protein expression level of the pathogenic gene, preferably the pathogenic gene being a gene overexpressed in tumors.

[0019] Furthermore, the application described in this invention is characterized by restoring the mutated C to U in the mRNA transcribed from a gene with a T-to-C pathogenic mutation, thereby enabling the expression of the pathogenic mutated gene to produce a protein with normal function.

[0020] Furthermore, the application described in this invention is characterized in that, based on the difference in substrate preferences between human APOBEC and ADAR, the APOBEC recruitment guide RNA (rgRNA) or RNA editing tool can simultaneously edit the target gene C-to-U and A-to-I.

[0021] Fourthly, the present invention provides the use of the APOBEC recruitment guide RNA (rgRNA) described in the first aspect or the RNA editing tool described in the second aspect in the preparation of medicaments for treating cancer or point mutation-related diseases.

[0022] Furthermore, the uses described in this invention are characterized in that the cancer overexpresses an oncoprotein, such as PIK3CA; the cancer is preferably breast cancer; and the point mutation-related diseases include cystic fibrosis caused by CRTR mutation, Usher syndrome caused by USH1C mutation, and human lupus caused by TLR7 mutation.

[0023] Fifthly, the present invention provides a composition for RNA editing, comprising: The APOBEC recruitment guide RNA (rgRNA) described in the first aspect or the RNA editing tool described in the second aspect; APOBEC protein, wherein the APOBEC protein is selected from APOBEC1 (A1), APOBEC3A (A3A), and APOBEC3G (A3G). And optionally, endogenous ADAR proteins.

[0024] Advantages and beneficial effects of the present invention: As a single RNA system, REVERSE achieves precise and flexible base switching through a "catch-recruit-execute" mechanism, expanding the boundaries of RNA editing in gene regulation and therapeutic applications. This has resulted in the following beneficial technical effects: First, the structure is simple and the editing efficiency is high.

[0025] rgRNAs generated through modular design, circularization strategies, and structural optimization can be delivered via plasmids, viral vectors, or chemically modified oligonucleotides; when a single cytidine conversion or simultaneous introduction of C-to-U and A-to-I mutations is used, the editing efficiency can reach up to approximately 60%.

[0026] Secondly, it has high safety and wide applicability, and can restore the function of disease-related mutated genes.

[0027] REVERSE exhibits high specificity across the transcriptome, with few global off-target events and limited bystander editing within the target region. REVERSE can function on a variety of endogenous mRNAs and non-coding RNAs in different cell types and can repair disease-related mutations without activating the innate immune response.

[0028] Third, it provides a new paradigm for cancer treatment.

[0029] Mounting evidence suggests that aberrant A3A is a key driver of tumor heterogeneity, metastasis, and drug resistance, and also an attractive but challenging therapeutic target. We repurposed aberrant A3A to introduce precise lethal mutations into oncogenic transcripts; in vivo delivery of a single rgRNA inhibited tumor growth, proposing a novel paradigm for cancer treatment. By redirecting APOBEC, involved in breast tumorigenesis, to key oncogenic transcripts, precise lethal mutations were introduced; in vivo delivery of circular rgRNA via lipid nanoparticles (LNPs) resulted in approximately 45% RNA editing of the mouse Pik3ca transcript and induced a reduction in tumor volume of approximately 65%. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0031] Figure 1 RNA editing using endogenous APOBEC (a) REVERSE design: The guide RNA contains a programmable antisense spacer to determine binding to the target RNA; it also contains two APOBEC recruitment domains to guide endogenous APOBECs to the hybridization region of the guide RNA and the target RNA. Site-directed C-to-U editing on the mRNA is controlled by the guide RNA, which is capable of inducing loop formation in the target RNA.

[0032] (b) Schematic diagram of the BFP-to-GFP reporter system construction.

[0033] (c) Representative images of rgRNA-induced GFP fluorescence in HEK293T cells that do not express or stably express APOBEC3A. NT indicates non-targeted; the naming convention for targeting rgRNAs is: number of recruitment domains, antisense spacer length, and edit loop length. Scale bar is 200 μm.

[0034] (d) Representative FACS analysis of rgRNA-induced GFP expression in HEK293T cells that do not express or stably express APOBEC3A.

[0035] (e) The bar chart shows that the proportion of rgRNA-induced GFP expression is correlated with APOBEC3A expression in HEK293T cells; n = 3, mean ± standard deviation.

[0036] (f) Representative western blot images showing APOBEC3A protein expression levels, including parental HEK293T cells, APOBEC3A knockdown cells, A3A 293T cells transfected with non-targeting siRNA, and APOBEC3A overexpressing cells. Vinculin was used as a loading control.

[0037] (g) Representative electrophoresis peak diagrams show the Sanger sequencing results after BG rgRNA or NT rgRNA induces the target region.

[0038] (h) The bar chart shows the C-to-U conversion rate of the target site as measured by NGS; n = 3, mean ± standard deviation.

[0039] Figure 2: RNA editing using specific APOBECs (a) A schematic diagram shows the construction of the No Stop (NS) and No Stop Bystander (NB) reporter systems used in this study. PTD stands for proteasome targeting domain, and the sequence is ALAPYIP.

[0040] (b) Bar charts show the effect of BG rgRNA or NT RNA on the expression level of targeted BG reporter transcripts in A3A 293T cells. P-values ​​were determined by unpaired two-tailed t-tests.

[0041] (cd) Representative images show rgRNA-induced GFP fluorescence in HEK293T cells that do not express or stably express APOBEC3G (c) or APOBEC1 (d). Scale bar 200 μm.

[0042] (ef) Bar charts show the C to U conversion rate at the target site in HEK293T cells stably expressing APOBEC3G(e) or APOBEC1(f). Data are from NGS results; n = 3, mean ± standard deviation.

[0043] Figure 3 Optimize rgRNA appropriately to improve efficiency and fidelity. (a) Schematic diagram showing the nucleotide frequency landscape of height-edited structural motifs at different stem lengths. Position numbering is relative to the reactive cytidine (0 site).

[0044] (b) Heatmaps show the editing efficiency of rgRNAs containing different recruiting domain (RD) variants; RDs are derived from high-confidence structural motifs and engineered through cyclic nucleotide alterations and stem base pair substitutions. Data are mean values ​​(n = 3).

[0045] (c) Representative images show GFP fluorescence in A3A293T cells after transfection with engineered rgRNAs containing optimal recruitment domains of different stem lengths. Scale bar 200 μm.

[0046] (d) Line graph showing the effect of antisense spacer length on RNA editing efficiency; n = 3, mean ± standard deviation.

[0047] (e) The bar chart shows the editing activity of REVERSE on all 16 triplet motifs in the NS reporter, with the nearest neighbor bases around the target cytidine being variable; n = 3, mean ± standard deviation.

[0048] (f) Line graph showing the effect of editing loop length on RNA editing efficiency; n = 3, mean ± standard deviation.

[0049] (g) The diagram above shows the NS reporter transcript sequence covered by the rgRNA, with target cytidine marked in red and bystander adenosine and cytidine marked in green and blue, respectively. The heatmap below shows the editing rate of engineered rgRNA variants on adenosine and cytidine covered in the NBreporter transcript, including the position within the induced editing loop (thick red box) and adjacent rgRNA binding regions (thick blue box). Data are mean (n = 3).

[0050] The editing efficiency in (b), (f), and (g) was determined by NGS.

[0051] Figure 4: Programmable negation spacer forms an editing loop within the target area. (a) The schematic diagram shows the formation of a 10nt editing loop after rgRNA captures the NS reporter transcript (left), and a representative on-target sequence (right) containing the target cytidine (red) and the tunable loop (black box).

[0052] (b) The schematic diagram shows the formation of a 10nt editing loop after rgRNA captures the NB reporter transcript (left), and a representative on-target sequence (right) containing the target cytidine (red), the bystander cytidine (blue), and the tunable loop (black box).

[0053] Figure 5 Chimeric design of antisense spacer-recruitment domains in rgRNA (a) Heatmap showing adenosine and cytidine editing rates induced by engineered rgRNA variants targeting NB reporter; data were measured by NGS and are presented as mean (n = 3).

[0054] (b) Schematic diagram showing the NS reporter transcript sequence covered by rgRNAVC (top) and the corresponding rgRNAVC design (bottom). RBS represents RNA binding sequence; AS represents antisense seed; eRD / A3A represents the engineered recruitment domain of APOBEC3A.

[0055] Figure 6 rgRNA circularization optimization (a) The diagram above illustrates the use of outward binding primers for RT-PCR to confirm rgRNA circularization; the diagram below shows the generation of crgRNA containing different RNA linkers in cells by gel electrophoresis, and is one of three repeated experiments.

[0056] (b) Representative images show GFP fluorescence in A3A-293T (top) and HeLa (bottom) cells after transfection with different rgRNA variants. Scale bar 200 μm.

[0057] (cd) Bar charts show the proportions of linear and circular rgRNA-containing different linkers induced by GFP expression in A3A-293T(c) and HeLa(d) cells based on NS, NB, and BG reporter data. n = 3, mean ± standard deviation.

[0058] Figure 7 Chemically modified circular rgRNA (a) The synthesis and circularization process of modified rgRNA.

[0059] (b) Representative images show GFP fluorescence in A3A-293T cells after transfection (top) or LNP delivery (bottom) of IVT crgRNA containing the modified nucleotides shown. The illustration is one of three replicates. Scale bar 200 μm.

[0060] (c) The schematic diagram shows that the 2,6 diaminopurine base (Z) and thymine (T) are fully paired via three intermolecular hydrogen bonds (dashed lines). Compared to adenine, the additional amino group at position 2 enhances its stability through the third hydrogen bond.

[0061] Figure 8 Editing endogenous transcripts in multiple cell lines using REVERSE (a) Schematic diagram showing circular rgRNA-mediated RNA editing achieved through multiple delivery strategies.

[0062] (b) Bar chart comparison of RNA editing efficiency achieved in TP53 transcripts by linear and circular rgRNA design via specific APOBEC recruitment; n = 3, mean ± standard deviation.

[0063] (c) Bar chart showing the inhibition of p53 transcriptional activation by RNA editing, with readings representing p53-driven luciferase expression; n = 3, mean ± standard deviation.

[0064] (dh) bar charts show the RNA editing efficiency achieved in endogenous transcripts (including coding and non-coding regions) of various cell lines via plasmid delivery, in vitro transcription (IVT), or lentiviral transduction of crgRNA. In Z-modified IVT crgRNA, adenosine in the antisense spacer is replaced by 2,6-diaminopurine ribonucleosides. n = 3, mean ± standard deviation.

[0065] (i) Bar chart comparing the RNA editing efficiency achieved by RESCUE-S and REVERSE in pathogenic transcripts; n = 3, mean ± standard deviation.

[0066] The editing efficiency in (b), (d), (e), (f), (g), (h) and (i) was determined by NGS.

[0067] (j) Representative images show clone formation experiments in BT 474 cells after expression of non-targeted or PIK3CA-targeted crgRNA.

[0068] Figure 9 Immunogenicity of genetically encoded and IVT crgRNA The heatmap shows the relative expression levels of type I interferon, interferon-stimulated genes, and pro-inflammatory genes (assessed by qPCR) after transfection with the indicated crgRNA and poly(I:C), compared with the untreated control. n = 3, mean ± standard deviation.

[0069] Figure 10 Effects of REVERSE on the expression levels of targeted endogenous transcripts (ae) Bar charts show changes in the expression levels of RNA editing-related target transcripts (assessed by qPCR) and compared with a non-target crgRNA control. P-values ​​were determined by unpaired two-tailed t-tests; NS indicates no significance, P>0.05; *P≤0.05. n = 3, mean ± standard deviation.

[0070] Figure 11: Reengineering crgRNA to achieve simultaneous adenosine and cytidine editing (a) The schematic diagram on the left shows the construction of a two-color reporter system for monitoring the two types of nucleotide transitions. The representative image on the right shows BFP and GFP fluorescence in A3A-293T cells after transfection with the reengineered crgRNA variant shown. The illustration shows one of three replicate experiments. Scale bar 200 μm.

[0071] (b) Schematic diagram showing the iterative optimization of crgRNAVS design.

[0072] (c) Bar chart showing the efficiency of simultaneous C-to-U and A-to-I editing on the Dp71b transcript in A3A-293T cells. Data were quantified by NGS, mean ± standard deviation (n = 3).

[0073] Figure 12 : Taming the abnormal APOBEC3A in breast cancer (a) Box plots show elevated APOBEC3A expression in the breast cancer cohort. Data were obtained from the TCGA database, and p-values ​​were determined by the Mann-Whitney U test.

[0074] (b) Schematic diagram showing the human PIK3CA transcript sequence covered by crgRNA, displaying the complementarity-binding region (pink shading), the editing loop (black box), and the target cytidine (red). C-to-U editing truncates the p110α protein by converting the sense codons of glutamine (Q) or arginine (R) within the red box to premature stop codons (X).

[0075] (c) Bar chart showing the efficiency of multiple C-to-U editing on the PIK3CA transcript in BT-474 cells. Data were quantified by NGS, mean ± standard deviation (n=3).

[0076] (d) Schematic diagram showing the mouse Pik3ca transcript sequence covered by crgRNA, showing the complementary binding region (pink shading), editing loop (black box) and target cytidine (red).

[0077] (e) Representative images show clone formation experiments in A3A-4T1 cells after expression of non-targeted or Pik3ca-targeted crgRNA.

[0078] (f) Bar charts show the efficiency of multiple C-to-U editing on the Pik3ca transcript in A3A-4T1 cells. Data were quantified by NGS, mean ± standard deviation (n = 3).

[0079] Figure 13 Transcriptome-wide RNA editing specificity of circular rgRNAs (a) RNA seq was used to analyze the effect of PPIB-targeting crgRNAs on gene expression at the transcriptome level. FPKM values ​​were calculated using the STRINGTIE tool, and Pearson correlation coefficient analysis was used to assess global differential gene expression.

[0080] (b) Analysis of the impact of PPIB-targeting crgRNA on natural editing sites based on RNA seq data. Pearson correlation coefficients were used to assess differences in RNA editing rates at natural editing sites. Four independent experiments were conducted.

[0081] (c) Bar chart comparing the off-target amounts of transcriptome range C-to-U and A-to-I induced by REVERSE and RESCUE S in A3A 293T cells.

[0082] The analyses shown in (a), (b) and (c) are all based on RNA data from four independent experiments.

[0083] Figure 14 Characterization of lipid nanoparticles (a) Representative transmission electron microscopy (TEM) images showing LNPs carrying non-targeting crgRNA (left) or crgRNA VSPik3ca (right). Scale bar 100 nm.

[0084] (b) Scatter plot showing the ζ-potential of LNPs loaded with non-targeting crgRNA versus crgRNA vs-Pik3ca; n = 3, mean ± standard error.

[0085] (c) Line plots show the hydration particle size distribution of LNPs loaded with non-targeted crgRNA and crgRNA vs. Pik3ca; the curves represent the mean ± standard error of three biological replicates.

[0086] Figure 15: Transcriptome changes associated with tumor regression after in vivo editing (a) GSEA figure shows the differential expression of PI3K-Aktsignaling pathway between combined therapy and non-targeted crgRNA treatment of mouse tumors.

[0087] (b) Volcano plot comparing gene expression in mouse tumors treated with combination therapy and non-targeted crgRNA. Labeled genes are those related to the PI3K-Akt signaling pathway. Data in (a) and (b) were obtained from n = 6 mice in each group.

[0088] (cd) KEGG enrichment analysis showed that, compared with the non-targeted crgRNA control, crgRNA vs. Pik3ca monotherapy (c) or combination therapy (d) treated mouse tumors showed the most significant differential regulation of the pathway.

[0089] Figure 16 REVERSE achieves efficient and specific RNA editing and induces tumor regression in mice. (a) A schematic diagram shows the delivery of IVT crgRNA LNP via intratumoral injection (it) and outlines the key steps in animal experiments to evaluate in vivo editing.

[0090] (b) Line graphs show the changes in mouse body weight in different groups throughout the experiment; n = 15 in each group before day 9 and n = 10 in each group after day 9. Data are mean ± standard deviation. P-values ​​were determined by one-way ANOVA.

[0091] (c) Line graphs show the average tumor growth curves for different groups; n = 15 per group before day 9, and n = 10 per group after day 9. Data are mean ± standard deviation. P-values ​​were determined by unpaired two-tailed t-tests.

[0092] (d) Image of the tumor harvested at the endpoint (left) and corresponding weight measurement (right). Scale bar 1 cm. n = 10 per group, mean ± standard deviation. P-value was determined by unpaired two-tailed t-test.

[0093] (e) Heatmap showing the C-to-U editing frequency of all cytidine sites in the target region of mouse tumors. Target cytidines are marked in red, and bystander cytidines are marked in blue. Data were quantified by NGS and are presented as the mean of three technical replicates.

[0094] (f) Bar chart showing the efficiency of multiple C-to-U editing on the Pik3ca transcript in mouse tumors. Data were quantified by NGS, mean ± standard deviation (n = 5).

[0095] (g) Bar charts show the changes in Pik3ca transcript expression levels compared to DPBS or non-targeted crgRNA controls. n = 10 per group, mean ± standard deviation. P-values ​​were determined by unpaired two-tailed t-tests.

[0096] (h) GSEA plot shows the differential expression of the PI3KAkt signaling pathway between crgRNA and Pik3ca and non-targeted crgRNA-treated mouse tumors.

[0097] (i) Volcano plot comparing gene expression in mouse tumors treated with crgRNA vs. Pik3ca and non-targeted crgRNA, with the labeled genes being those related to the PI3K Akt signaling pathway.

[0098] (h) and (i) data are from each group of n = 6 mice.

[0099] (jk) Representative Western blot images (j) and quantitative results (k) show the p110α levels in tumors of mice in different treatment groups; GAPDH was used as a loading control. n = 9 per group, mean ± standard deviation. P-values ​​were determined by unpaired two-tailed t-tests.

[0100] (l) Representative immunohistochemical images showing staining of p110α, Ki 67 and PCNA (proliferation), cleaved caspase 3 and TUNEL assay (apoptosis) in mouse tumor tissue sections after different treatment interventions. Scale bar 50 μm.

[0101] Figure 17 Safety assessment of LNP-crgRNA monotherapy and combination therapy in mice (a) Tables show the results of complete blood count analysis at the study endpoint. P-values ​​were determined by univariate ANOVA to compare DPBS with LNP-crgRNA or combination therapy groups. P < 0.05 was considered statistically significant. NS indicates P > 0.05; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

[0102] (b) Representative HE-stained images showing histological analysis of major organs after the indicated treatment intervention. Scale bar 100 μm.

[0103] Figure 18 We designed rgRNAs containing the APOBEC recruitment region for TP53 Q192X, R213X, R196X, and Q52X.

[0104] Figure 19 crgRNAs that target the 3'UTR or coding sequences of KRAS, SMARCA4, PPIB, CTNNB1, TYMS, and FTO transcripts.

[0105] Figure 20 crgRNAs that target the 3'UTR or coding sequences of DKC, RAB7A, RAF1, NRAS, and EZH2 transcripts.

[0106] Figure 21 crgRNAs targeting MALAT1 and XIST transcripts Figure 22 REVERSE demonstrates its therapeutic potential for correcting disease-related T to C point mutations by targeting crgRNAs of CFTR L453S, CFTR L558S, USH1C F743S, TLR7 Y264H, and Dp71b Y118H+R122Q.

[0107] Figure 23 Multiple PTC crgRNAs were simultaneously introduced into human PIK3CA and mouse PIK3CA transcripts. Detailed Implementation

[0108] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0109] The experimental materials and methods used in the embodiments of this invention are as follows:

[0110] 1. Plasmid construction For the linear rgRNA expression construct, the synthesized rgRNA sequence was cloned into the pXR003 gRNA cloning backbone (Addgene no. 109053) via Gibson assembly, and rgRNA transcription was driven by the hU6 promoter. For the genetically encoded crgRNA expression construct, the inventors constructed a cloning vector based on the pXR003 backbone, containing Twister P3 U2A, 5' ligation sequence, 5' linker sequence, 3' linker sequence, 3' ligation sequence, and Twister P1, and then cloned the rgRNA sequence into this autocatalytic circular RNA expression vector using Gibson assembly. For the lentiviral crgRNA expression construct, the autocatalytic crgRNA sequence with the upstream hU6 promoter was amplified by PCR and cloned into the pLVX-puro vector (Clontech). For in vitro transcription of crgRNA, the inventors first cloned the group I self-splicing intron and homologous arms into a PCR linearized plasmid vector containing a T7 RNA polymerase promoter, and then replaced the region between the homologous arms with rgRNA and linker sequences.

[0111] To construct the BFP-GFP fluorescence reporter (BG reporter), the inventors added a Y66H mutation (TAC to CAC) or a Y66H and W57X combination mutation (TGG to TAG) to the GFP cassette via PCR, and cloned the PCR product into the pcDNA3.1(+) vector using Gibson assembly. To construct two versions of the EGFP fluorescence reporter (NS and NB reporter), the inventors amplified the EGFP coding sequence without a stop codon by PCR, then fused it with a protein linker sequence (containing an in-frame CGA codon) and a proteasome targeting peptide sequence via overlap-PCR, and subsequently cloned it into pcDNA3.1(+). Bystander cytidine sites, target triplet motifs, and other minor modifications were introduced via mutagenesis PCR.

[0112] To achieve stable expression of human APOBEC3A, APOBEC3G, or APOBEC1 in cell lines, the coding sequences for each deaminase were synthesized and cloned into the pSIH1 H1 Puro vector. An upstream minimal CMV promoter provided low-level, continuous transcription to mimic physiological expression conditions. To regulate A3A expression, the RfxCas13d gene was amplified from the pXR001 EF1aCasRx 2A EGFP plasmid and cloned into pcDNA3.1(+). To compare editing efficiency and off-target effects, RESCUE-S (pCMV-dCas13b6-mapkNES-GS-dADAR2 V16S) was purchased from Addgene.

[0113] To construct vectors expressing disease-related mutant genes, the inventors synthesized and amplified the full-length coding sequences of the Dp71b transcript variants of CFTR, USH1C, TLR7, and DMD. Subsequently, T-to-C and G-to-A mutations were introduced via mutagenesis PCR according to ClinVar definitions. The amplified products were cloned into the pLVX puro vector using Gibson assembly.

[0114] 2. Cell Culture and Cell Line Construction All cells were cultured at 37 °C and 5% CO2. HEK293T, HepG2, and HT-29 cells were cultured in DMEM with 10% FBS and 1% penicillin streptomycin. HeLa cells were cultured in MEM with 10% FBS and 1% P / S. MCF7 cells were cultured in DMEM with 10% FBS, 1% P / S, and 0.01 mg / mL insulin. BT474 cells were cultured in RPMI 1640 with 20% FBS, 1% P / S, 0.01 mg / mL insulin, and 2 mM Lglutamine. 4T1 cells were cultured in RPMI 1640 with 10% FBS and 1% P / S. All cell lines were negative for mycoplasma and identified by STR.

[0115] To establish a cell line stably expressing a specific deaminase, the inventors transfected HEK293T cells with the A3A / A3G / A1 construct based on the pSIH1 H1 Puro backbone and the packaging plasmids pRSV Rev, pMDLg / pRRE, and pMD2.G using PEI. After 72 h, the viral supernatant was collected, concentrated using PEG, and stored at 80°C. HEK293T or 4T1 cells were infected with lentivirus and treated with puromycin for 7 days. Single clones were isolated from puromycin-resistant cells, amplified, preserved, and verified by sequencing and Western blot.

[0116] 3. In vitro preparation of circular RNA The preparation of crgRNA involved modifications to existing methods. In short, the crgRNA precursor was obtained from a linearized crgRNA template via in vitro transcription using a T7 High Yield RNA Synthesis Kit. After IVT, the product was treated with DNase I for 30 min to digest the DNA template. To generate chemically modified crgRNA, the corresponding modified NTP was completely replaced with a final concentration of 10 mM during IVT rgRNA template digestion; this was followed by DNase I digestion, and three RNA fragments—3' intron RNA, modified rgRNA, and 5' intron RNA—were ligated via splint-mediated ligation. To catalyze crgRNA cyclization, a final concentration of 2 mM GTP was added to the reaction, and the mixture was heated at 55 °C for 15 min. The RNA product was then purified using a Monarch RNA Cleanup Kit. To enrich crgRNA, column-purified RNA was diluted with RNase-free water, heated at 65 °C for 3 min, cooled on ice for 3 min, and then incubated with RNase R and 10×RNase R buffer at 37 °C for 15 min. The RNase R-digested RNA was then purified again by column purification.

[0117] 4. Exogenous transcript editing To measure RNA editing using a fluorescence reporter system, A3A 293T cells were seeded into 6-well plates 24 hours before transfection, with approximately 6 × 10⁶ cells per well. 5 Cells were transfected. When cell confluence reached approximately 70%, 2 μg of linear or circular rgRNA plasmids were transfected according to the PEI instructions, and 1 μg of reporter plasmid was co-transfected. Forty-eight hours after transfection, editing efficiency was assessed quantitatively by cell imaging and the proportion of GFP-positive cells. In FACS analysis, BFP signal was used as a fluorescent selection marker for reporter-expressing cells, and the percentage of GFP+ / BFP+ cells was calculated as the editing efficiency reading.

[0118] 5. Endogenous transcript editing To evaluate RNA editing on endogenous mRNA and lncRNA transcripts, the inventors used approximately 6 × 10⁶ A3A-293T, HeLa, Hep G2, or 4T1 cells per well. 5 Approximately 3 × 10⁶ HT 29 cells per well. 5 Approximately 1 × 10⁶ BT-474 cells are present per well.6 4 μg of circular rgRNA plasmid was transfected into a 6-well plate at approximately 70% confluence using PEI. Forty-eight hours after transfection, total RNA was extracted using the FastPure Cell / Tissue Total RNA Isolation Kit V2 and processed using HiScript IV 1. st The Strand cDNA Synthesis Kit reverse transcribed the DNA into cDNA. For IVT crgRNA, cells were transfected with 120 pmol IVT crgRNA and 12 μl Lipofectamine 3000 when approximately 50% confluence was reached; cells were harvested after 24 h, RNA was extracted, and reverse transcribed. For lentiviral crgRNA, crgRNA was generated and lentivirus was delivered by transfecting HEK293T cells with the pLVX-crgRNA-puro transgenic plasmid and packaging plasmids psPAX2 and pMD2.G; the concentrated virus was then used to infect Hep G2 or HT-29 cells at an MOI of approximately 10. Cells were collected after 72 h for RNA extraction and cDNA preparation. Subsequently, PCR was performed using appropriate primers to amplify a region of approximately 300-600 bp surrounding the target site, with the amplified region located outside the antisense domain length. The cycle number was tested to ensure linear amplification. After PCR products were subjected to gel electrophoresis, the target size bands were purified using a FastPure Cell / Tissue DNA Isolation Mini Kit column for Sanger sequencing or Illumina NGS.

[0119] 6. Tumor cell clonogenesis assay After digestion and resuspending, cells were seeded at 1,000 cells per well into six-well plates 24 h after lentiviral transduction (BT-474) or 8 h after transfection (4T1). After culturing in complete medium at 37 °C for 7–10 days, tumor cells were gently washed twice with PBS and fixed with 4% paraformaldehyde for 30 min. After crystal violet staining for 30 min, imaging was performed using the IncuCyteSX5 system.

[0120] 7. Preparation and characterization of crgRNA-loaded LNPs Lipid nanoparticles were prepared using a microfluidic system via ethanol dilution. In short, ionizable cationic lipids SM-102, DSPC, cholesterol, and DMG-PEG2000 were gently mixed in an ethanol phase at a molar ratio of 50:10:38.5:1.5 with gentle pipetting, resulting in a total volume of 1.0 mL. Subsequently, using a Nanoassemblr and Hamilton liquid handling system, crgRNA LNPs were prepared by mixing 3 parts by volume of IVT crgRNA with 1 part by volume of lipid solution, where the crgRNA:lipids weight ratio was 1:20, resulting in a final lipid concentration of 2.81 mg / mL (4.80 mM). The crude formulation was purified and buffer-replaced by overnight dialyzing in 1,000 volumes of DPBS using 10K MWCO Slide A LyzerG2 dialysis cassettes. To characterize the quality of crgRNA LNPs, TEM was used to observe the size and morphology of the nanoparticles; at the same time, the hydrated particle size (ζ-size) and surface charge (ζ-potential) were measured using a Nano ZSZetasizer at a backscattering angle of 173°.

[0121] 8. Animal experiments BALB / c mice were purchased from Vital River and housed in the animal facility of Peking University School of Medicine under conditions of 20-25℃, 40-70% humidity, controlled light from 08:00 to 20:00, SPF conditions, and free access to food and water. Six-week-old BALB / c mice were anesthetized with 2% isoflurane and then subcutaneously injected with 1×10⁻⁶ ppm of isoflurane into the fat pad of the pre-shaved right mammary gland. 6 A3A-4T1 tumor cells (100 μl DPBS). When the tumor volume reaches 100-200 mm. 3 On day 7, tumor-bearing mice were randomly divided into five groups. They were then injected with DPBS, non-targeting crgRNA-LNP (0.5 mg / kg), and other treatments, respectively. -1 ), crgRNAVS-Pik3ca-LNP (0.5mg·kg -1 Nab-PTX (10 mg·kg) -1 (Or combined treatment) Two days after the first treatment, five mice in each group were sacrificed to collect tumor tissue to assess editing efficiency. Mice were monitored every two days during the experiment. The length and width of the tumor were measured twice a week using calipers, and the tumor volume was calculated as V = 1 / 2 × length × width. 2Calculations were performed. At the study endpoint (day 17), no tumors reached the ethical termination threshold (diameter exceeding 2 cm in any direction). All mice were anesthetized, and blood samples were collected from the orbital sinus for complete blood count analysis. Simultaneously, tumor tissue was excised, weighed, and imaged, and divided into three portions: one portion was ground in liquid nitrogen for protein extraction and target degradation measurement; another portion was homogenized in 1 mL TRIzol for RNA extraction and transcriptome analysis; and the third portion was fixed in 4% PFA for paraffin embedding. Subsequently, major organs were collected and fixed in 4% PFA for histological safety assessment. All animal experiments were approved by the Experimental Animal Management Committee of Peking University Health Science Center.

[0122] 9. Quantitative Analysis of Target RNA Editing In Sanger sequencing, RNA editing efficiency was quantified as the peak height ratio T / (C+T). In NGS analysis, reads were demultiplexed and indexed to the target site sequence using Cutadapt. Alignment and quantification were performed using Bowtie / RSEM; BAM was sorted using SAMtools, and RNA editing sites were analyzed using Reditools with specified parameters. Fisher's exact test was used to identify significant C-to-T and / or A-to-G transitions within the rgRNA target region with a p-value < 0.05, considered as rgRNA-induced editing events.

[0123] 10. Global off-target analysis of the transcriptome mRNA fractions were enriched using the NEBNext Poly(A) mRNA Magnetic Isolation Module, followed by library construction using the NEBNext Ultra RNA Library Prep Kit for Illumina. The libraries were sequenced on an Illumina HiSeqXten PE150, with approximately 20 million reads per sample. To exclude non-specific effects caused by transfection reagents, mock groups were set up, processing only the transfection reagents. Each group contained four replicates. Reads were aligned to the human reference genome GRCh38 hg38 using STAR software in two rounds. Repeated reads were removed using Picard, and SNVs were identified using GATKHaplotypeCaller and filtered according to criteria including QD>2, sequencing coverage>20×, and wild-type samples supporting reference allele reads>99%. Figure 4, c shows the edit sites as the intersection of the four replicates.

[0124] 11. Transcriptome-wide RNA seq analysis Tissue RNA was extracted using chloroform extraction, followed by reverse transcription, PCR amplification, and NGS sequencing (PE150, 30G per sample). The bioinformatics workflow followed the work of Vogel et al. and Qu et al. FastQC was used for quality control, and Cutadapt was used for quality cuts: the first 6 bp of each read was cut, with a maximum cut of 20 bp. Reads shorter than 90 nt after cut were filtered. The filtered reads were then aligned to the mouse reference genome GRCm39 mm39 using STAR. The inventors used GATK HaplotypeCaller to retrieve variants. The raw VCF file generated by GATK was filtered and annotated using GATK VariantFiltration, bcftools, and ANNOVAR. Variants in dbSNP, 1000 Genomes, and EVS were filtered out. Four duplicate-shared variants in each group were then selected as RNA editing sites.

[0125] 12. Quantitative PCR Each 20 μl qPCR reaction was mixed with 1 μl cDNA using Hieff UNICON Universal SYBR Master Mix, and primers were designed to amplify fragments shorter than 300 bp. All samples were performed in triplicate on a LightCycler 480 II. Transcript levels were measured using GAPDH as an internal control and quantified according to the 2-ΔΔCt method.

[0126] 13. Western blot After harvesting cells or processing tissues, lysis was performed using 1×RIPA buffer supplemented with a complete protease inhibitor cocktail. Cell debris was removed by centrifugation at 16,000g for 10 min at 4°C. Proteins were extracted with lysis buffer and quantified using the BCA assay. 100 μg of protein from each sample was boiled with loading buffer, separated by 10% SDS-PAGE, and transferred to a 0.22 μm PVDF membrane. The membrane was blocked in TBST with 5% skim milk at room temperature for 1 h, washed three times with TBST, and then incubated sequentially with primary and secondary antibodies. Protein bands were detected using a BeyoECL Moon and visualized on a Tanon 5200 imaging system. Band density was quantified using ImageJ. Antibodies used included anti-APOBEC3A, anti-PI3K p110α, anti-GAPDH, anti-Vinculin, and HRP-labeled secondary antibodies for detecting mouse or rabbit primary antibodies.

[0127] 14. Complete blood count analysis Blood samples collected from the orbital sinus of mice were placed in tubes containing K2-EDTA and stored at 4 °C. Hematological parameters were analyzed using a BC-2800Vet Auto Hematology Analyzer.

[0128] 15. Immunohistochemistry Tumor tissues were fixed with 4% PFA and then dehydrated sequentially with 70%, 95%, and 100% ethanol, followed by defatting with xylene for 2 h and then paraffin embedding. FFPE tumor tissues were sectioned into 10 μm sections using a Leica RM2125 RTS microtome, dewaxed, rehydrated, and then subjected to antigen retrieval. Endogenous peroxidase activity was quenched by incubation with 3% hydrogen peroxide at room temperature for 25 min. After blocking with 5% normal goat serum in PBS at room temperature for 1 h, the sections were incubated overnight at 4 °C with the corresponding primary antibodies: anti-PI3K p110α, anti-Ki67, anti-PCNA, and anti-cleaved caspase-3. The TUNEL assay was performed using the In Situ Cell Death Detection Kit according to the instructions to detect apoptotic cells. In short, after dewaxing and rehydration, the sections were permeabilized with proteinase K, then the TUNEL reaction mixture was added, and incubated in a humidified chamber at 37 °C for 1 h. After washing, the sections were incubated with species-matched HRP-labeled secondary antibody at room temperature for 50 min, followed by DAB staining and hematoxylin counterstaining. After dehydration, clearing, and mounting, images were acquired using a WS 10 pathology slide scanner and analyzed using NDP.view2 software.

[0129] 16. Histological analysis 10 μm FFPE sections were prepared from major organs including the heart, liver, spleen, lungs, and kidneys. After standard HE staining, the sections were dehydrated, cleared, mounted, and whole-section imaging was performed on a WS-10 pathology scanner. Quantitative and qualitative analyses were performed using NDP.view2 software.

[0130] 17. Statistics and Repeatability The number of biological replicates in each experiment is denoted by n. All graphs and statistical analyses were performed using GraphPadPrism 8.0.2. Unpaired two-tailed Student's t-tests were used for comparisons between two groups; one-way ANOVA was used for comparisons of three or more groups, with pairwise comparisons corrected for multiple comparisons.

[0131] Example 1: RNA editing using endogenous APOBEC Among the reported cytidine deaminases, APOBEC1 (A1), APOBEC3A (A3A), and APOBEC3G (A3G) can all target sequences with stem-loop structures in human RNA. Based on this, the inventors envisioned generating a guide RNA that can both recruit endogenous APOBECs and specify RNA targets to form editing loops.

[0132] To verify this hypothesis, the inventors first established a HEK293T-derived cell line that stably expresses A3A at low levels, named A3A-293T, and constructed a BFP-to-GFP reporter (BG reporter) to evaluate C-to-U editing efficiency. The trifunctional guide RNA, named rgRNA, consists of two parts: first, two SDHB hairpin loops modified from the A3A target site characterized in SDHB mRNA, serving as recruitment domains; and second, a 40 nt antisense spacer that is inversely complementary to BG mRNA, capable of inducing a tetranucleotide loop containing the target cytidine.

[0133] Following transfection of hU6 promoter-driven rgRNA into A3A 293T cells, RNA editing-inducible fluorescence conversion was observed. The absence of sequence-matching rgRNA, lack of endogenous A3A expression, or both failed to effectively activate GFP expression. Furthermore, knockdown of A3A using CRISPR / CasRx eliminated rgRNA-induced GFP signaling; and with increasing A3A dosage, the proportion of GFP-positive cells increased from 18% to approximately 40%, indicating that the A3A protein is responsible for rgRNA-induced fluorescence conversion. Sanger sequencing revealed C / U overlap at the predicted cytidine site of the CAC codon, indicating a significant C-to-U conversion. No bystander editing of other cytidines or endogenous ADAR-mediated A-to-I(G) conversion was detected significantly within the target region. NGS further confirmed that the C-to-U editing efficiency was approximately 11%. Figure 1 qPCR analysis also ruled out the possibility that rgRNA reduced the level of the target transcript through RNAi effects.

[0134] To determine whether rgRNA could be engineered to recruit other cytidine deaminases, the inventors replaced the recruitment domain with either the PAPSAP2 hairpin loop or the ApoB hairpin loop, and observed robust GFP expression and editing levels in A3G-293T or A1-293T, respectively. These results demonstrate that rgRNA can induce RNA editing on target transcripts by recruiting specific APOBEC proteins. Figure 2 ).

[0135] Example 2: Characterization and Optimization of REVERSE The recruitment domain is a key component of rgRNA and functions to increase the local concentration of APOBEC protein. The inventors investigated several properties of the recruitment domain by screening variants in parallel. The recruitment domain is derived from highly edited A3A target sites in the human transcriptome, most of which have predictable secondary structures, typically consisting of a NAUC tetranucleotide loop and palindromic arms of varying lengths (2-9 nt per arm).

[0136] Systematic analysis of these palindromic sequences revealed a length-related nucleotide pattern. Short palindromes (2–4 nt) primarily form rigid paired stems through simple C / G repeats; longer motifs (5–9 nt) contain a centrally conserved adenosine, which may serve as a tunable structural hinge; and 7–9 nt motifs also contain a uridine-rich 3' end fragment, which may further fine-tune stem-loop stability.

[0137] Based on the nucleotide pattern of the natural A3A substrate, the inventors constructed a series of recruitment domains with different stem lengths and evaluated their effectiveness using a BG reporter. The results showed that a 5 bp stem yielded the highest editing output, and the editing efficiency was significantly dependent on stem length. Further modification of the tetranucleotide loop-3 site revealed that CAUC and UAUC were favorable motifs for optimization.

[0138] Because the stem sequences of natural A3A substrates also exhibit diversity, the inventors subsequently constructed a library of recruitment domain variants containing computationally predicted, highly reliable base pair substitutions. The highest editing efficiency came from a combination of a CAUC tetranucleotide loop and an AU-to-GC substitution at position -6 in a 5 bp stem; this structure was named eRD / A3A (engineered recruiting domain for A3A). Specific base pair substitutions also improved editing efficiency at other stem lengths, but none surpassed eRD / A3A, further demonstrating that stem length is the primary determinant of recruitment domain function.

[0139] Subsequently, the inventors titrated the antisense spacer length from 20 nt to 200 nt to determine the optimal placement and design of the rgRNA containing eRD / A3A. To better evaluate rgRNA activity, the inventors constructed an eGFP No Stopreporter (NS reporter) by fusing the proteasome targeting domain PTD (ALAPYIP) to the C-terminus of an enhanced GFP; the deamination of the target cytidine introduces a premature stop codon, preventing PTD translation and thus avoiding eGFP degradation. Based on the BGreporter and NS reporter, the inventors found that the antisense spacer length is positively correlated with editing efficiency, with an optimal length of approximately 100-150 nt. Figure 3 ).

[0140] The inventors also tested the targeting flexibility of REVERSE. For all 16 triad combinations (5'-N1CN2-3') on the NS reporter, the antisense spacer length was fixed at 100 nt, ensuring the formation of preferred editing loops. NGS results showed that A3A activity was highest when the position preceding C was U, and it tolerated changes in the UCN2 motif; A1 and A3G preferred the ACN2 and CCN2 motifs, respectively. The complementary specificity of different APOBECs allows REVERSE to expand the range of editable targets by recruiting different endogenous APOBECs.

[0141] Example 3: Rational design of editing loop to improve fidelity The inventors hypothesized that altering the editing loop might improve editing efficiency, given that APOBEC-mediated deamination requires specific substrate structure. To this end, they designed a series of rgRNAs targeting two different reporter transcripts (NS reporter and NB reporter) to form editing loops of varying lengths. Figure 4 NGS results showed that target cytidines in smaller rings (less than 8 nt) could be edited with moderate efficiency; excitingly, the editing efficiency of both reporters increased with increasing ring size, peaking at 16 nt; larger rings (greater than 24 nt) reduced or eliminated C-to-U conversion. This suggests that 16 nt rings may be more suitable for target deamination than tetranucleotide rings. Figure 3 ).

[0142] Because APOBEC tends to nonselectively deaminate multiple cytidines on RNA bulges or single-stranded RNA, the inventors examined all cytidine sites within the antisense spacer coverage region of the target NB transcript. The results showed that the edited bystander sites were located in single-stranded RNA sequences adjacent to the edit loop or coverage region, consistent with APOBEC's preference for ssRNA substrates. While no adenosine deamination was observed in the 40 nt rgRNA, several A-to-I(G) transitions were observed in the 100 nt coverage sequence, as longer rgRNA-transcript duplexes may become substrates for endogenous ADAR1 / 2.

[0143] By adjusting the length of the editing loop, the inventors can exclude some bystander sites (e.g., -5, -20, and +18) from the loop, thereby minimizing bystander editing within the loop; however, other off-target effects still exist and affect RNA editing specificity.

[0144] Based on these findings, the inventors proposed that, given the differences in substrate profiles between APOBEC and ADAR, this type of off-target effect can be eliminated by masking bystander cytidine with the double-stranded RNA region while simultaneously preventing editable adenosine from being exposed to the double-stranded substrate. The inventors reserved a 40 nt central antisense domain as an antisense seed (AS) to match the target transcript and form a tetranucleotide loop, presenting only the target cytosine to the catalytic pocket. Instead of simply extending the antisense region, the inventors designed separate mRNA binding sequences (RBSs) covering multiple regions of the target mRNA containing highly editable cytidine but no adenosine.

[0145] The antisense seed and RBS together constitute a flexible, large antisense spacer, interspersed with engineered recruitment domains. This design was envisioned to synergistically satisfy both the need for strong binding of target transcripts and efficient recruitment of deaminases. Subsequently, the inventors optimized this design for an rgRNA targeting the NB reporter, termed NB-rgRNAVC (a chimeric antisense spacer recruiting domain version), resulting in significantly improved on-target editing efficiency while eliminating or minimizing off-target editing within the sequencing window. Figure 5 ).

[0146] Example 4: Circular rgRNA Enables Efficient RNA Editing Circularizing the target RNA improves its stability, enabling more persistent protein expression and stronger A-to-I RNA editing. To further enhance rgRNA-mediated C-to-U editing, the inventors used an autocatalytic Twisterribozyme to generate genetically encoded circular rgRNAs (crgRNAs) and evaluated their effects by targeting the aforementioned reporter.

[0147] Although covalent circularization of crgRNA was detected, its editing efficiency was roughly equivalent to that of linear rgRNA on NS and NB reporters; it even decreased on BG reporter. The inventors speculate that certain structures of the guide RNA, especially the recruitment domain, may be perturbed by the circularization process, thereby interfering with APOBEC recruitment and target recognition.

[0148] Therefore, the inventors placed flexible RNA linkers on both sides of the linear rgRNA to reduce the adverse effects of circularization on the guide RNA structure. Comparison of four different RNA linkers revealed that adding the AC linker increased the editing efficiency of the three reporters by up to approximately 2 times compared to the corresponding linear forms. In HeLa cells, crgRNA containing the flexible linker also exhibited higher editing efficiency. In summary, the inventors optimized the construction method for circularizing rgRNA; adding the AC linker resulted in more persistent guide RNA and improved RNA editing efficiency. Figure 6 ).

[0149] Example 5: Editing endogenous transcripts in multiple cell lines using REVERSE To confirm the ability and universality of editing endogenous RNA in the form of crgRNA, the inventors chose the tumor suppressor gene TP53 as a target. TP53 is a core gene for maintaining genome stability and regulating gene expression. The inventors systematically designed a set of rgRNAs that induce C-to-U mutations in the mRNAs of TP53 by redirecting A3A, A3G, and A1, respectively. For the Q192X and R213X mutations of TP53, rgRNAs containing the APOBEC3A (A3A) recruitment region were designed; for the R196X mutation of TP53, rgRNAs containing the APOBEC3G (A3G) recruitment region were designed; and for the Q52X mutation of TP53, rgRNAs containing the APOBEC1 (A1) recruitment region were designed (see [link to documentation]). Figure 18 ).

[0150] Figure 18The sequences for the APOBEC3A (A3A) recruitment regions of Q192X and R213X, as shown in gccatctatcgatggg (SEQ ID NO:1), are used for precise editing of the Q192 and R213 sites; the sequences for the APOBEC3G (A3G) recruitment regions of R196X, as shown in gattctgcccagaaga (SEQ ID NO:2), are used for precise editing of the R196 site; and the sequences for the APOBEC1 (A1) recruitment regions of Q52X, as shown in catatatgatacaatttgatcagtatatt (SEQ ID NO:3), are used for precise editing of the Q52 site.

[0151] These rgRNAs effectively recruit specific APOBECs to edit each target cytidine, and by introducing early stop codons into TP53 mRNA, they inhibited p53 transcriptional regulatory activity by up to approximately 4-fold. Results based on the p53-luciferasecis-reporting system showed that the circular form with the AC linker was superior to the linear form. This provides preliminary evidence that crgRNA-mediated endogenous RNA editing is robust.

[0152] Subsequently, the inventors designed crgRNA targeting 11 transcripts [(KRAS, SMARCA4, PPIB, CTNNB1, TYMS, FTO, see...]]. Figure 19 (DKC, RAB7A, RAF1, NRAS, and EZH2, see also) Figure 20 The 3'UTR or coding sequence of the RNA, and two non-coding RNAs (MALAT1 and XIST, see [link]). Figure 21 The study was conducted in multiple cell types expressing endogenous A3A. NGS analysis confirmed that plasmid delivery of genetically encoded crgRNA successfully edited target cytidine in A3A-293T and MCF7 cells.

[0153] The sequences of the APOBEC3A (A3A) recruitment regions for the above 13 transcripts are shown in ggatgctatcgcatcg (SEQ ID NO:4).

[0154] Furthermore, the inventors observed that in vitro transcribed (IVT) crgRNA generated using group I intron-mediated autocatalysis retained its function after delivery via LNP. Given the importance of chemical modification in advancing nucleic acid drugs, the inventors also explored whether modifying nucleotides in IVT crgRNA could improve editing efficiency. The results showed that replacing adenine with 2,6-diaminopurine (Z) in the antisense spacer improved editing rates in both the BG reporter and endogenous genes, possibly because ZU pairing forms more intermolecular hydrogen bonds, increasing the double-strand stability of the crgRNA target RNA. Conversely, substitutions with Ψ, N1-methylpseudouridine (Z) (N1mΨ), or 5-methylcytidine (m5C), which suppress innate immune responses, did not significantly differ from unmodified crgRNA in target editing, suggesting that the potential immunogenicity of crgRNA may not be a major obstacle affecting editing performance. Figure 7 ).

[0155] The genetically encoded crgRNA is also compatible with viral delivery, such as lentiviruses. In difficult-to-transfect cells like Hep G2 and HT-29, lentiviral delivery can produce higher levels of editing. Similarly, the inventors reengineered the crgRNA to recruit endogenous A3G or A1, and achieved target RNA editing of eight additional transcripts in A3G-293T or A1-293T, respectively. Figure 8 ).

[0156] Unlike dsRNA poly(I:C), neither genetically encoded crgRNA nor IVT crgRNA triggered innate immune activation; the expression of type I interferon and pro-inflammatory genes did not change significantly. Figure 9 Furthermore, the inventors did not observe crgRNA-mediated knockdown of endogenous mRNA or non-coding RNA. Figure 10 ).

[0157] Example 6: Correcting pathogenic mutations and inhibiting oncoprotein expression using structure-optimized crgRNA The inventors further explored the therapeutic potential of REVERSE for correcting disease-related T to C point mutations. Results showed that REVERSE could restore pathogenic mutations in the CFTR (cystic fibrosis), USH1C (Usher syndrome), and TLR7 (human lupus) transcripts, respectively. (See [link to relevant documentation]). Figure 22Its efficiency is comparable to that of the catalytically inactivated CRISPR Cas13 (dCas13) basal RNA editor. Figure 8 i).

[0158] The APOBEC recruitment region sequence in the crgRNAs of CFTR L453S, CFTR L558S, USH1C F743S, and TLR7 Y264H is shown in taggcatatctgcctt (SEQ ID NO:5). The APOBEC recruitment region sequence in the crgRNA of Dp71b Y118H+R122Q is shown as gagcatatctgcctggagtcggaggcatatctgcctggatgaggaggcatatctgcctgacaggcgaggcatatctgcctgccaccggaggcatatctgcctggtgagggaggcatatctgcctgcgtcgagaggcatatctgcctggcctaggaggcatatctgcctggtaccgtc (SEQ ID NO:6). ADAR recruitment region sequence such as gacggtacaaactaggcaaatcgacgaaacctcacaaacggtggaaagcctgtaaactcatcaaacgactcaaacacactggccgggcgcggtggctc acgcctgtaatcccagcactttggggggccgaggcggggagattgcttgagcccaggagttcgagaccagcctgggcaacatagcgagaccccgtctcagtgtg (seq ID NO:7) or as shown in gtgtgaagccgggcgtggtggcgcgcgcctgtagtcccagctactcgggaggctgaggcaggaggatcgcttgagcccaggagttcgaggctgcagtgagctatgatcgcgccactgcactccagcctgggcgacagagcgagaccctgtctctcacac (SEQ ID NO:8) shown.

[0159] As more and more studies have found that multinucleotide variations (MNVs) are associated with genetic diseases, the inventors envisioned that a crgRNA capable of simultaneously converting cytidine and adenosine in the same transcript would significantly expand REVERSE's RNA editing capabilities. To this end, the inventors reconstructed the BG reporter, using blue fluorescence recovery to monitor A-to-I editing and green fluorescence conversion to assess simultaneous editing.

[0160] Using a long antisense guide RNA to recruit the circular form of endogenous ADAR, the inventors found robust blue fluorescence recovery at the central A-Cmismatch, indicating high adenosine conversion efficiency, consistent with previous studies. Coupled with the A3Arecruiting domain to the long antisense guide RNA, moderate cytidine conversion was achieved within the induced editing loop; however, the addition of the GluR2 domain, derived from a natural ADAR substrate, did not further enhance editing efficiency. Compared to the aforementioned circular guide RNAs, the circular chimeric rgRNA (crgRNAVC) showed similar efficiency in A-to-I editing and superior performance in C-to-U editing.

[0161] Despite the effectiveness of the system, the inventors further optimized the crgRNA design to improve editing efficiency. The optimization ideas include: (1) retaining the flexible antisense spacer (40nt antisense seed and 6×20 nt RBS) to maintain high-affinity target binding and specifically mask bystander sites; (2) introducing a multiple deaminase recruitment domain array to enhance the editing of both types of bases; and (3) replacing some of the A3A recruiting domains in the flexible antisense spacer with a 3×A linker to minimize potential interference with the crgRNA structure.

[0162] In fluorescent reporter and human dystrophin Dp71b transcripts, optimized circular rgRNAs binding to Alu elements improved A-to-I editing efficiency. Notably, the A3Arecruiting domain of minimal duet arrays achieved C-to-U editing efficiency comparable to crgRNAVC; the highest efficiency was achieved with octet arrays or more arrays. Based on these findings, the inventors defined an optimal circular rgRNA design, named crgRNAVS. This design can efficiently and faithfully correct two pathogenic mutations in the Dp71b transcript simultaneously (… Figure 11Genomic analysis of patient tumors revealed that APOBEC3 cytidine deaminase is a key driver of mutation formation in tumor evolution and drug resistance. Since C-to-U RNA editing can convert sense codons into premature stop codons, such as converting the CAG glutamine codon to the amber stop codon UAG, the inventors hypothesized that REVERSE could utilize A3A in malignant tumors with aberrant A3A expression and elevated activity to install PTCs in oncogenic transcripts, thereby truncating oncoproteins.

[0163] In human BT-474 breast cancer cells that stably express A3A, optimal crgRNAVS design can simultaneously introduce multiple PTCs into the PIK3CA transcript encoding p110α. Figure 23 p110α is the catalytic subunit of PI3K and a core signaling hub controlling tumor cell proliferation and survival pathways. Targeting PTC installation strongly inhibits tumor cell growth, suggesting that utilizing tumor-intrinsic A3A to inactivate key oncogenes may represent a potential therapeutic strategy. Figure 8 ,j; Figure 12 ). Among them, the APOBEC recruitment region sequence is such as gaggcatatctgcctggagtcggaggcatatctgcctggatgaggaggcatatctgcctgacaggcgaggcatatctgcctgccaccggaggcatatctgcctggtgagggaggcatatctgcctgcgtcgagaggcatatctgcctggcctaggaggcatatctgcctggtaccgtc (SEQ ID NO:6) shown; ADAR recruitment region sequence such as gacggtacaaactaggcaaatcgacgaaacctcacaaacggtggaaagcctgtaaactcatcaaacgactcaaacacactggccgggcgcggtggctc acgcctgtaatcccagcactttggggggccgaggcggggagattgcttgagcccaggagttcgagaccagcctgggcaacatagcgagaccccgtctcagtgtg (seq ID NO:7) or as shown in gtgtgaagccgggcgtggtggcgcgcgcctgtagtcccagctactcgggaggctgaggcaggaggatcgcttgagcccaggagttcgaggctgcagtgagctatgatcgcgccactgcactccagcctgggcgacagagcgagaccctgtctctcacac (SEQ ID NO:8) shown.

[0164] Example 7: Transcriptome-wide specificity of crgRNA To assess the effects of REVERSE on cellular gene expression and normal RNA editing, the inventors used transcriptome-wide RNA sequencing (RNA-seq) to compare the transcriptome characteristics of PPIB-targeted circular rgRNA samples with those of untransfected A3A-293T samples. The results showed on-target editing in the circular rgRNA samples; correlation analysis indicated that crgRNA-mediated editing had no significant impact on the global transcriptome, including the expression levels of the target PPIB transcript, consistent with previous RT-qPCR results.

[0165] Furthermore, the global C-to-U and A-to-I RNA editing sites shared by the PPIB-targeted circular rgRNA group and the untransfected A3A-293T control group were highly consistent, suggesting that crgRNA had minimal perturbation to the normal RNA editing map. Unlike the nearly 430 off-target edits in the transcriptome caused by exogenous editor RESCUE-S overexpression, the circular rgRNA transfection group showed only a few potential off-targets: 23 undesirable C-to-U sites and 22 A-to-I sites, respectively. Figure 13 ).

[0166] Example 8: In vivo editing of REVERSE

[0167] Given that crgRNAVS significantly improves RNA editing efficiency and fidelity, the inventors further tested whether REVERSE could mediate intracorporeal RNA editing in the context of cancer treatment. Unlike humans, who have seven different A3 enzymes, mice encode only one homologous protein, APOBEC3, which has low catalytic activity but can inhibit reverse transcription during viral infection.

[0168] Therefore, the inventors first established a mouse breast cancer cell model (A3A-4T1) capable of expressing human A3A enzymes, and designed crgRNAVS to introduce PTCs (Q96X, R115X, and R140X) into the mouse Pik3ca transcript, corresponding to the studies in the human cell model. crgRNAVS-Pik3ca clearly mediated multiple editing in the A3A 4T1 cell model and correspondingly inhibited tumor cell growth.

[0169] To achieve in vivo delivery of crgRNAVS, the inventors prepared lipid nanoparticles using ionizable amino lipid SM-102, which encapsulated IVT crgRNAVS-Pik3ca, and confirmed that they formed uniformly sized solid spherical particles. Figure 14 Subsequently, the inventors evaluated the antitumor efficacy of crgRNAVS-Pik3ca encapsulated in A3A-4T1 tumor-bearing mice by intratumoral injection every 7 days at a dose of 0.5 mg kg−1 LNP.

[0170] The combination therapy with crgRNA versus Pik3ca produced significant antitumor activity, inhibiting tumor growth by approximately 65%, comparable to the FDA-approved albumin-bound paclitaxel nanoformulation Nab-PTX (administered intravenously every 4 days). Notably, the combination therapy achieved a tumor volume reduction of approximately 90% at the study endpoint, demonstrating the clinical compatibility of LNP-delivered crgRNA with conventional cancer treatment.

[0171] Two days after the initial treatment, 5 out of 15 mice in each group were sacrificed and tumors were collected to detect RNA editing events. Editing was detected only in tumors injected with crgRNAVS-Pik3ca; it was not detected in paclitaxel monotherapy, DPBS, or the non-targeted crgRNAVS control. Both crgRNAVS-Pik3ca monotherapy and combination therapy mediated efficient C-to-U conversion at the target cytidine, with mean editing efficiencies of approximately 31% / 27%, 21% / 19%, and 15% / 14% at the Q96X, R115X, and R140X sites, respectively, and no bystander editing was detected in vivo.

[0172] Notably, up to approximately 45% of Pik3ca transcripts underwent RNA editing at one or more target sites, with about 9% simultaneously carrying three intended PTCs. The frequencies were consistent across the two treatment groups, indicating the robustness of synchronous RNA editing in vivo.

[0173] Since premature stop codons can lead to truncated protein production or mRNA degradation via nonsense-mediated mRNA decay (NMD), the inventors further inquired whether targeted PTC installation would trigger Pik3ca transcript decay and impair p110α protein synthesis. At the study endpoint, C to U editing did not cause a statistically significant change in Pik3ca transcript levels compared to untreated mice, although a decreasing trend was observed. Transcriptome analysis also supported no significant change in edited mRNA abundance, but revealed significant inhibition of the downstream PI3K AKT signaling pathway, consistent with anti-tumor efficacy. Figure 15 ).

[0174] Subsequently, the inventors measured the p110α protein level in the harvested tumors and found that, compared with paclitaxel monotherapy or the non-targeted control, only LNP-encapsulated crgRNA vs. Pik3ca treatment resulted in a 37% reduction in p110α expression in mice, indicating that targeted PTC installation successfully truncated part of the oncoprotein. Immunohistochemistry further confirmed that the tumor proliferation markers Ki67 and PCNA were reduced in both the crgRNA vs. Pik3ca monotherapy and combination therapy groups, while apoptosis pathway activation (increased Caspase 3 and TUNEL assay) was also observed. Figure 16 Furthermore, all treated mice showed normal weight gain, no adverse reactions, comparable blood counts to the untreated control, and no significant abnormalities in the histology of major organs, indicating that REVERSE has good in vivo safety. Figure 17 ).

[0175] In summary, LNP delivery of crgRNA can redirect endogenous APOBEC, introduce an early stop codon into the oncogenic transcript, and demonstrates promising safety and anti-tumor efficacy in vivo.

[0176] This invention defines a new generation of transcriptome engineering tools by utilizing the ability of a single guide RNA to achieve site-directed RNA editing. Such tools complement exogenous editor methods and possess greater translational potential due to the development paradigms of clinically validated nucleic acid drugs such as siRNA and ASO. The rgRNA generated by this invention expands the capabilities of single-RNA site-directed RNA editing systems, enabling not only specific A-to-I or C-to-U editing, but also simultaneous adenosine and cytidine conversion and multiple base editing, and on-demand utilization of endogenous ADAR and / or APOBEC deaminases. Because rgRNA is a modular, programmable molecular device, REVERSE can continue to improve its functionality and versatility in several ways.

[0177] First, the inventors obtained a series of recruitment domains from natural, highly reliable APOBEC substrates through computational-aided functional screening. In the future, methods such as systematic evolution of non-repressive RNA aptamers can be used to explore a broader sequence space, generating fully synthetic recruitment domains with higher affinity.

[0178] Second, for the IVT form, studies have shown that introducing 2,6-diaminopurine into the antisense spacer can increase editing yield, suggesting that more RNA base modifications could be systematically explored to further improve performance.

[0179] Third, one of the key functions of rgRNA is to induce tunable editing loops. Studies have shown that loop lengths of 4-20 nt are typically functional, allowing for user-defined single-base or multi-adjacent-base editing by adjusting the loop's position and size. Notably, this observation is consistent with the tailored crRNA-induced editing loop phenomenon observed in dCas13 basal RNA editor studies. Integrating these complementary results will provide preliminary insights into the substrate binding and catalytic mechanisms of natural and evolved ssRNA deaminases.

[0180] Fourth, the inventors developed a flexible antisense spacer design that eliminates bystander editing while maintaining on-target efficiency. This strategy is achieved by strategically placing unwanted cytidines within base-pairing regions and keeping unwanted adenosines outside these regions. Further combining this with U deletion, G·U wobble base pair, or MIRROR strategies could potentially further improve RNA editing precision and efficiency.

[0181] Fifth, circularized guide RNAs have been used in previous studies and in this study to improve stability and target residence time, thereby enhancing RNA editing results. However, the cumulative effects of crgRNAs, especially long-term accumulation in vivo, still require careful monitoring.

[0182] Sixth, APOBEC exhibits sequence bias, making certain cytidines unfavorable for editing. Consistent with recent biochemical and structural analyses, APOBEC shows strong editing inhibition of the 5'-GCN-3' motif preceded by G, and also exhibits lower editing of motifs followed by U or C, such as the reduced editing of UCC and UCU motifs by A3A. A deeper understanding of APOBEC deaminase activity, especially the high-resolution structure of the APOBEC-RNA complex, will facilitate the rational design of rgRNAs, addressing steric hindrance issues and enhancing editing at unfavorable motifs.

[0183] To date, RNA editing methods have primarily focused on correcting pathogenic mutations and manipulating protein function. This study proposes a novel application of REVERSE in cancer therapy: repurposing the A3A mutation, a major driver of tumor evolution, to introduce precise lethal mutations at the RNA level to suppress tumors. Data show that LNP delivery of IVT crgRNA is sufficient to achieve efficient multiplex base editing and achieves antitumor effects comparable to the commonly used chemotherapy drug Nab PTX in tumor-bearing mice. Because REVERSE is inherently simple, specific, scalable, and compatible with evolving tumor-specific RNA delivery systems, rgRNA-based therapy holds immense potential in oncology, reflecting a shift in the tumor-targeted therapy paradigm from "inhibiting overactivated oncogenic proteins" to "repurposing abnormally active proteins to treat tumors."

[0184] Finally, the inventors summarized a "capture-recruitment execution" model for developing endogenous RNA-modifying enzymes for site-specific RNA editing. Unlike existing A-to-I or U-to-Ψ RNA editing tools that rely on guide RNA-target transcript duplexes to mimic natural dsRNA substrates, rgRNA represents a universal design utilizing diverse enzymes acting on ssRNA: it presents substrates by inducing loop structures in the target region. With over 170 RNA modifications and their regulatory mechanisms systematically studied, modular rgRNAs can be used for programmable transcriptome engineering with specific enzymes by replacing recruitment domains, adjusting loop lengths, optimizing antisense spacers for specific targets, or customizing expression cassettes (e.g., Pol II promoters, small nuclear RNA frameworks). Further development of such tools is expected to unlock the ability to rewrite epitranscriptome information, reshape RNA-interactomes, and regulate protein translation through RNA-based site-specific editing. Overall, REVERSE not only enables precise and flexible RNA editing with broad applications, but also provides a simple and flexible guide RNA design for utilizing endogenous RNA modifying enzymes, potentially facilitating RNA center research and drug development.

[0185] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. The patent protection scope of the present invention should be determined by the described claims.

[0186] The following prior art is incorporated herein by reference in its entirety: 1. Nishikura, K. A to I editing of coding and non-coding RNAs by ADARs. Nat Rev Mol Cell Biol 17, 83–96 (2016). 2. Eisenberg, E.&Levanon, EY A to I RNA editing—immune protectorand transcriptome di versifier. Nat Rev Genet 19, 473–490 (2018). 3. Pecori, R., Di Giorgio, S., Paulo Lorenzo, J.&Nina Papavasiliou,F. Functions and conse quences of AID / APOBEC mediated DNA and RNAdeamination. Nat Rev Genet 23, 505–518 (2022). 4. Cervantes Gracia, K., Gramalla Schmitz, A., Weischedel, J.&Chahwan, R. APOBECs orchestrate genomic and epigenomic editing across healthand disease. Trends in Genetics 37, 1028–1043(2021). 5. Tan, M. H. et al. Dynamic landscape and regulation of RNA editingin mammals. Nature 550, 249–254 (2017). 6. Gagnidze, K., Rayon Estrada, V., Harroch, S., Bulloch, K.&Papavasiliou, F. N. A New Chapter in Genetic Medicine: RNA Editing and itsRole in Disease Pathogenesis. Trends in Molecular Medicine 24, 294–303(2018). 7. Levanon, E. Y. et al. Systematic identification of abundant A to Iediting sites in the human transcriptome. Nat Biotechnol 22, 1001–1005(2004). 8. Nishikura, K. Functions and Regulation of RNA Editing by ADARDeaminases. Annual Review of Biochemistry 79, 321–349 (2010). 9. Savva, Y. A., Rieder, L. E.&Reenan, R. A. The ADAR protein family.Genome Biol 13, 252 (2012). 10. Salter, J. D.&Smith, H. C. Modeling the Embrace of a Mutator:APOBEC Selection of Nucleic Acid Ligands. Trends in Biochemical Sciences 43,606–622 (2018). 11. Salter, J. D., Bennett, R. P.&Smith, H. C. The APOBEC ProteinFamily: United by Structure, Divergent in Function. Trends in BiochemicalSciences 41, 578–594 (2016). 12. Rees, H. A.&Liu, D. R. Base editing: precision chemistry on thegenome and transcriptome of living cells. Nat Rev Genet 19, 770–788 (2018). 13. Pfeiffer, L. S.&Stafforst, T. Precision RNA base editing withengineered and endogenous effectors. Nat Biotechnol 41, 1526–1542 (2023). 14. Khosravi, H. M.&Jantsch, M. F. Site directed RNA editing: recentadvances and open challenges. RNA Biology 18, 41–50 (2021). 15. Marina, R. J., Brannan, K. W., Dong, K. D., Yee, B. A.&Yeo, G. W.Evaluation of Engineered CRISPR Cas Mediated Systems for Site Specific RNAEditing. Cell Reports 33, (2020). 16. Cox, D. B. T. et al. RNA editing with CRISPR Cas13. Science 358,1019–1027 (2017). 17. Abudayyeh, O. O. et al. A cytosine deaminase for programmablesingle base RNA editing. Science 365, 382–386 (2019). 18. Huang, X. et al. Programmable C‐to‐U RNA editing using the humanAPOBEC 3A deaminase. The EMBO Journal 39, e104741 (2020). 19. Liu, Z., Jillette, N., Robson, P.&Cheng, A. W. Simultaneousmultifunctional transcriptome engineering by CRISPR RNA scaffold. NucleicAcids Research 51, e77–e77 (2023). 20. Stafforst, T.&Schneider, M. F. An RNA–Deaminase ConjugateSelectively Repairs Point Mutations. Angewandte Chemie International Edition51, 11166–11169 (2012). 21. Vogel, P. et al. Efficient and precise editing of endogenoustranscripts with SNAP tagged ADARs. Nat Methods 15, 535–538 (2018). 22. Latifi, N., Mack, A. M., Tellioglu, I., Di Giorgio, S.&Stafforst,T. Precise and efficient C to U RNA base editing with SNAP CDAR S. NucleicAcids Research 51, e84–e84 (2023). 23. Stroppel, A. S. et al. Harnessing self labeling enzymes forselective and concurrent A to I and C to U RNA base editing. Nucleic AcidsResearch 49, e95–e95 (2021). 24. Montiel Gonzalez, M. F., Vallecillo Viejo, I., Yudowski, G. A.&Rosenthal, J. J. C. Correction of mutations within the cystic fibrosistransmembrane conductance regulator by site directed RNA editing. Proceedingsof the National Academy of Sciences 110, 18285–18290 (2013). 25. Sinnamon, J. R. et al. In Vivo Repair of a Protein Underlying aNeurological Disorder by Programmable RNA Editing. Cell Reports 32, 107878(2020). 26. Katrekar, D. et al. In vivo RNA editing of point mutations viaRNA guided adenosine deaminases. Nat Methods 16, 239–242 (2019). 27. Rauch, S. et al. Programmable RNA Guided RNA Effector ProteinsBuilt from Human Parts. Cell 178, 122 134.e12 (2019). 28. Han, W. et al. Programmable RNA base editing with a single gRNAfree enzyme. Nucleic Acids Research 50, 9580–9595 (2022). 29. Ichinose, M. et al. U to C RNA editing by synthetic PPR DYWproteins in bacteria and human culture cells. Commun Biol 5, 968 (2022). 30. Vallecillo Viejo, I. C., Liscovitch Brauer, N., Montiel Gonzalez,M. F., Eisenberg, E.&Rosen thal, J. J. C. Abundant off target edits from sitedirected RNA editing can be reduced by nuclear localization of the editingenzyme. RNA Biology 15, 104–114 (2018). 31. Charlesworth, C. T. et al. Identification of preexisting adaptiveimmunity to Cas9 proteins in humans. Nat Med 25, 249–254 (2019). 32. Tang, X. Z. E., Tan, S. X., Hoon, S.&Yeo, G. W. Pre existingadaptive immunity to the RNA editing enzyme Cas13d in humans. Nat Med 28,1372–1376 (2022). 33. Teoh, P. J. et al. Aberrant hyperediting of the myelomatranscriptome by ADAR1 confers oncogenicity and is a marker of poorprognosis. Blood 132, 1304–1317 (2018). 34. Law, E. K. et al. APOBEC3A catalyzes mutation and drivescarcinogenesis in vivo. Journal of Experimental Medicine 217, e20200261(2020). 35. Chew, W. L. et al. A multifunctional AAV–CRISPR–Cas9 and its hostresponse. Nat Methods 13, 868–874 (2016). 36. Merkle, T. et al. Precise RNA editing by recruiting endogenousADARs with antisense oligonucleotides. Nat Biotechnol 37, 133–138 (2019). 37. Qu, L. et al. Programmable RNA editing by recruiting endogenousADAR using engineered RNAs. Nat Biotechnol 37, 1059–1069 (2019). 38. Monian, P. et al. Endogenous ADAR mediated RNA editing in nonhuman primates using stereopure chemically modified oligonucleotides. NatBiotechnol 40, 1093–1102 (2022). 39. Reautschnig, P. et al. CLUSTER guide RNAs enable precise andefficient RNA editing with endogenous ADAR enzymes in vivo. Nat Biotechnol40, 759–768 (2022). 40. Bass, B. L.&Weintraub, H. An unwinding activity that covalentlymodifies its double stranded RNA substrate. Cell 55, 1089–1098 (1988). 41. Woolf, T. M., Chase, J. M.&Stinchcomb, D. T. Toward thetherapeutic editing of mutated RNA sequences. Proceedings of the NationalAcademy of Sciences 92, 8298–8302 (1995). 42. Tian, N. et al. A structural determinant required for RNAediting. Nucleic Acids Res 39, 5669–5681 (2011). 43. Matthews, M. M. et al. Structures of human ADAR2 bound to dsRNAreveal base flipping mechanism and basis for site selectivity. Nat Struct MolBiol 23, 426–433 (2016). 44. Powell, L. M. et al. A novel form of tissue specific RNAprocessing produces apolipoprotein B48 in intestine. Cell 50, 831–840 (1987). 45. Wolfe, A. D., Li, S., Goedderz, C.&Chen, X. S. The structure ofAPOBEC1 and insights into its RNA and DNA substrate selectivity. NAR Cancer2, zcaa027 (2020). 46. Kouno, T. et al. Crystal structure of APOBEC3A bound to singlestranded DNA reveals structural basis for cytidine deamination andspecificity. Nat Commun 8, 15024 (2017). 47. Yang, H., Kim, K., Li, S., Pacheco, J.&Chen, X. S. Structuralbasis of sequence specific RNA recognition by the antiviral factor APOBEC3G.Nat Commun 13, 7498 (2022). 48. Petljak, M. et al. Characterizing Mutational Signatures in HumanCancer Cell Lines Reveals Episodic APOBEC Mutagenesis. Cell 176, 12821294.e20 (2019). 49. Petljak, M. et al. Mechanisms of APOBEC3 mutagenesis in humancancer cells. Nature 607, 799–807 (2022). 50. Angus, L. et al. The genomic landscape of metastatic breastcancer highlights changes in mutation and signature frequencies. Nat Genet51, 1450–1458 (2019). 51. Isozaki, H. et al. Therapy induced APOBEC3A drives evolution ofpersistent cancer cells. Nature 620, 393–401 (2023). 52. Rosenberg, B. R., Hamilton, C. E., Mwangi, M. M., Dewell, S.&Papavasiliou, F. N. Transcriptome wide sequencing reveals numerous APOBEC1mRNA editing targets in transcript 3’-UTRs. Nat Struct Mol Biol 18, 230–236(2011). 53. Sharma, S. et al. APOBEC3A cytidine deaminase induces RNA editingin monocytes and macrophages. Nat Commun 6, 6881 (2015). 54. Sharma, S. et al. Mitochondrial hypoxic stress induces widespreadRNA editing by APOBEC3G in natural killer cells. Genome Biology 20, 37(2019). 55. Standage Beier, K. et al. A transient reporter for editingenrichment (TREE) in human cells. Nucleic Acids Res 47, e120 (2019). 56. Sharma, S.&Baysal, B. E. Stem loop structure preference for sitespecific RNA editing by APOBEC3A and APOBEC3G. PeerJ 5, e4136 (2017). 57. Sharma, S., Patnaik, S. K., Taggart, R. T.&Baysal, B. E. Thedouble domain cytidine deaminase APOBEC3G is a cellular site specific RNAediting enzyme. Sci Rep 6, 39100 (2016). 58. Sharma, S., Patnaik, S. K., Kemer, Z.&Baysal, B. E. Transientoverexpression of exogenous APOBEC3A causes C to U RNA editing of thousandsof genes. RNA Biology 14, 603–610 (2017). 59. Nakata, Y. et al. Cellular APOBEC3A deaminase drives mutations inthe SARS CoV 2 genome. Nucleic Acids Research 51, 783–795 (2023). 60. Schneekloth, J. S. Jr. et al. Chemical Genetic Control of ProteinLevels: Selective in Vivo Targeted Degradation. J. Am. Chem. Soc. 126, 3748–3754 (2004). 61. Kim, K., Shi, A. B., Kelley, K.&Chen, X. S. Unraveling the EnzymeSubstrate Properties for APOBEC3A Mediated RNA Editing. Journal of MolecularBiology 435, 168198 (2023). 62. Barka, A. et al. The Base Editing Enzyme APOBEC3A CatalyzesCytosine Deamination in RNA with Low Proficiency and High Selectivity. ACSChem. Biol. 17, 629–636 (2022). 63. Wesselhoeft, R. A., Kowalski, P. S.,&Daniel G. Anderson.Engineering circular RNA for potent and stable translation in eukaryoticcells. Nat Commun 9, 2629 (2018). 64. Yi, Z. et al. Engineered circular ADAR recruiting RNAs increasethe efficiency and fidelity of RNA editing in vitro and in vivo. NatBiotechnol 40, 946–955 (2022). 65. Katrekar, D. et al. Efficient in vitro and in vivo RNA editingvia recruitment of endogenous ADARs using circular guide RNAs. Nat Biotechnol40, 938–945 (2022). 66. Litke, J. L.&Jaffrey, S. R. Highly efficient expression ofcircular RNA aptamers in cells using autocatalytic transcripts. NatBiotechnol 37, 667–675 (2019). 67. el Deiry, W. S. et al. WAF1, a potential mediator of p53 tumorsuppression. Cell 75, 817–825(1993). 68. Zhou, Y. et al. A widespread pathway for substitution of adenineby diaminopurine in phage genomes. Science 372, 512–516 (2021). 69. Karikó, K., Buckstein, M., Ni, H.&Weissman, D. Suppression of RNARecognition by Toll like Receptors: The Impact of Nucleoside Modification andthe Evolutionary Origin of RNA. Immunity 23, 165–175 (2005). 70. Wang, Q. et al. Landscape of multi nucleotide variants in 125,748human exomes and 15,708 genomes. Nat Commun 11, 2539 (2020). 71. Buisson, R. et al. Passenger hotspot mutations in cancer drivenby APOBEC3A and mesoscale genomic features. Science 364, eaaw2872 (2019). 72. Langenbucher, A. et al. An extended APOBEC3A mutation signaturein cancer. Nat Commun 12, 1602 (2021). 73. Gupta, A. et al. APOBEC3 mutagenesis drives therapy resistance inbreast cancer. Nat Genet 57, 1452–1462 (2025). 74. MacMillan, A. L., Kohli, R. M.&Ross, S. R. APOBEC3 Inhibition ofMouse Mammary Tumor Virus Infection: the Role of Cytidine Deamination versusInhibition of Reverse Transcription. Journal of Virology 87, 4808–4817(2013). 75. Kurosaki, T., Popp, M. W.&Maquat, L. E. Quality and quantitycontrol of gene expression by nonsense mediated mRNA decay. Nat Rev Mol CellBiol 20, 406–420 (2019). 76. Zhu, Y.&Hart, G. W. Dual specificity RNA aptamers enablemanipulation of target specific O GlcNAcylation and unveil functions of OGlcNAc on β catenin. Cell 186, 428 445.e27 (2023). 77. Pfeiffer, L. S. et al. Stereo random oligonucleotides enableefficient recruitment of ADAR in vitro and in vivo. Nat Commun 16, 8849(2025). 78. Zhuang, Y. et al. Single strand deaminase assisted editing forfunctional RNA manipulation. Nat Biotechnol 1–13 (2026) doi:10.1038 / s41587025 02956 7. 79. Reautschnig, P. et al. Precise in vivo RNA base editing with awobble enhanced circular CLUSTER guide RNA. Nat Biotechnol 43, 545–557(2025). 80. Sun, Y. et al. Improved RNA base editing with guide RNAsmimicking highly edited endogenous ADAR substrates. Nat Biotechnol 1–13(2025) doi:10.1038 / s41587 025 02628 6. 81. Doherty, E. E. et al. Rational Design of RNA Editing GuideStrands: Cytidine Analogs at the Orphan Position. J. Am. Chem. Soc. 143,6865–6876 (2021). 82. Doherty, E. E. et al. ADAR activation by inducing a synconformation at guanosine adjacent to an editing site. Nucleic Acids Research50, 10857–10868 (2022). 83. Gomerdinger, V. F., Nabar, N.&Hammond, P. T. Advancingengineering design strategies for targeted cancer nanomedicine. Nat RevCancer 25, 657–683 (2025). 84. Song, J. et al. CRISPR free, programmable RNA pseudouridylationto suppress premature termination codons. Molecular Cell 83, 139 155.e9(2023). 85. Adachi, H. et al. Targeted pseudouridylation: An approach forsuppressing nonsense mutations in disease genes. Molecular Cell 83, 637651.e9 (2023). 86. Sordyl, D. et al. MODOMICS: a database of RNA modifications andrelated information. 2025 update and 20th anniversary. Nucleic Acids Res 54,D219–D225 (2026). 87. Byrne, S. M. et al. An engineered U7 small nuclear RNA scaffoldgreatly increases ADAR mediated programmable RNA base editing. Nat Commun 16,4860 (2025). 88. Smargon, A. A. et al. Enhancing RNA base editing on mammaliantranscripts with small nuclear RNAs. Nat Chem Biol 1–9 (2025) doi:10.1038 / s41589 025 02026 8。

Claims

1. An APOBEC recruitment guide RNA (rgRNA), comprising: At least one APOBEC recruitment domain that can bind to the APOBEC protein; and At least one programmable antisense region, each programmable antisense region including an operably linked first antisense region and a second antisense region, wherein the first antisense region and the second antisense region are complementary to the first binding region and the second binding region of the target RNA sequence, respectively; When rgRNA binds to target RNA, a circular single-stranded editing loop is formed between the first and second binding regions on the target RNA sequence. The editing loop has a 5'-N1CN2-3' motif, where N1 and N2 are any ribonucleotides and cytosine C is the base editing site.

2. The APOBEC recruitment guide RNA (rgRNA) as described in claim 1, characterized in that, The APOBEC recruitment domain includes a stem-loop secondary structure formed by a single-chain loop and a palindromic arm, and is either an APOBEC recruitment domain of a natural APOBEC substrate or an artificially modified APOBEC recruitment domain based on a natural APOBEC substrate.

3. The APOBEC recruitment guide RNA (rgRNA) as described in claim 2, characterized in that, The APOBEC recruitment domain is a variant artificially modified based on the APOBEC recruitment domain of the natural APOBEC substrate; The length of the palindromic arm is 2-9 bp, and the single-stranded ring is an NAUC tetranucleotide ring, where N is any ribonucleotide; Preferably, the length of the palindromic arm is 5 bp, and the single-chain loop is selected from CAUC and UAUC; Preferably, the APOBEC recruitment domain has one or more structural units selected from SEQ ID NO:1-5, or has one or more structural units formed by deletion, substitution, or addition based on SEQ ID NO:1-5.

4. The APOBEC recruitment guide RNA (rgRNA) as described in claim 1, characterized in that, The length of the programmable antisense region is 20-200 nt, preferably 100-150 nt, and more preferably 150 nt; the adenine in the programmable antisense region can be replaced with 2,6-diaminopurine.

5. The APOBEC recruitment guide RNA (rgRNA) as described in claim 1, characterized in that, The APOBEC protein is selected from APOBEC1 (A1), APOBEC3A (A3A), and APOBEC3G (A3G). Preferred, When the editing ring has a 5'-ACN2-3' motif, the APOBEC recruitment structure domain combined with APOBEC1 (A1) is used; When the edit ring has a 5'-UCN2-3' motif, the APOBEC recruitment structure domain combined with APOBEC3A (A3A) is used; When the edit ring has a 5'-CCN2-3' motif, the APOBEC recruitment structure domain combined with APOBEC3G (A3G) is used.

6. The APOBEC recruitment guide RNA (rgRNA) as described in claim 1, characterized in that, The editing ring is a 4-24nt cyclic single chain, preferably a 16nt cyclic single chain.

7. The APOBEC recruitment guide RNA (rgRNA) as described in claim 1, characterized in that, It may also include one or more mRNA binding sequences (RBS), which bind to regions containing editable cytosine outside the first and second binding regions of the target RNA sequence to reduce off-target editing of the target RNA by the APOBEC protein on the extracyclic cytosine.

8. The APOBEC recruitment guide RNA (rgRNA) as described in claim 7, characterized in that, The RBS and the programmable antisense region have an APOBEC recruitment structure field; the multiple RBSs have an APOBEC recruitment structure field.

9. The APOBEC recruitment guide RNA (rgRNA) as described in claim 1, characterized in that, It is either linear rgRNA or circular rgRNA; preferably circular rgRNA, wherein the circular rgRNA is covalently circularized RNA after placing flexible RNA linkers on both sides of the linear rgRNA.

10. An RNA editing tool comprising, or capable of generating, the APOBEC recruiting guide RNA (rgRNA) according to any one of claims 1-9. Preferably, the RNA editing tool is selected from plasmids, LNP-rgRNA, and viral vectors; More preferably, the plasmid or viral vector can be transcribed into rgRNA within the cell, and the LNP-rgRNA can deliver rgRNA into the cell.

11. The RNA editing tool of claim 10, further comprising an ADAR recruitment domain, the ADAR recruitment domain being capable of recruiting endogenous ADAR and achieving A-to-I editing of the target gene; Preferably, the sequence of the ADAR recruitment domain is selected from SEQ ID NO:7-8.

12. The use of the APOBEC recruitment guide RNA (rgRNA) according to any one of claims 1-9 or the RNA editing tool according to any one of claims 10-11 in correcting pathogenic gene mutations and / or inhibiting the expression of pathogenic genes, wherein, The application does not alter the transcript levels of the target gene.

13. The application as described in claim 12, characterized in that, PTC is generated by introducing C-to-U into the mRNA sequence of the pathogenic gene, thereby reducing the protein expression level of the pathogenic gene. Preferably, the pathogenic gene is a gene overexpressed in tumors.

14. The application as described in claim 12, characterized in that, By restoring the mutated C to U in the mRNA transcribed from a gene with a T-to-C pathogenic mutation, the expression of the pathogenic gene can produce a protein with normal function.

15. The application as described in any one of claims 12-14, characterized in that, Based on the differences in substrate preferences between human APOBEC and ADAR, the APOBEC recruitment guide RNA (rgRNA) or RNA editing tool can simultaneously edit target genes C-to-U and A-to-I.

16. Use of the APOBEC recruiting guide RNA (rgRNA) of any one of claims 1-9 or the RNA editing tool of any one of claims 10-11 in the preparation of a medicament for treating cancer or point mutation-related diseases.

17. The use as described in claim 13, characterized in that, The cancer overexpresses an oncoprotein, such as PIK3CA; the cancer is preferably breast cancer; the point mutation-related diseases include cystic fibrosis caused by CRTR mutation, Usher syndrome caused by USH1C mutation, and human lupus caused by TLR7 mutation.

18. A composition for RNA editing, comprising: The APOBEC recruitment guide RNA (rgRNA) according to any one of claims 1-9 or the RNA editing tool according to any one of claims 10-11; Endogenous APOBEC proteins, wherein the APOBEC proteins are selected from APOBEC1 (A1), APOBEC3A (A3A), and APOBEC3G (A3G). Optional, endogenous ADAR protein.

Images