Development and application of high-efficiency and high-fidelity adenine base editor

Abstract

The invention provides development and application of a high-efficiency and high-fidelity adenine base editor. The inventor compares the activities of several adenine base editors (ABE) on a large number of sgRNA target sites, and discovers that ABE8e shows the highest editing efficiency and a wider editing window; however, a serious sgRNA independent DNA off-target effect is introduced into the ABE8e. Therefore, the inventor performs optimization and modification on ABE8e, reveals some sites related to off-target effect and editing accuracy, and obtains a series of TadA8e variants, and the variants significantly reduce the off-target effect of an adenine base editor and improve the editing accuracy.

Description

Technical Field

[0001] The present invention belongs to the field of genetic engineering technology; more specifically, the present invention relates to the development and application of adenine base editors based on the CRISPR system. Background Art

[0002] Single-base editors can mutate a single base at the target site without generating double-strand breaks (DSBs) in the genome, thereby avoiding chromosomal mismatches, ectopy, and recombination caused by double-strand breaks.

[0003] Cytosine base editors (CBEs) and adenine base editors (ABEs) can mediate mutations from C to T and A to G, respectively. Since about 58% of known genetic diseases are caused by single-base mutations, and ABEs can treat 47% of single-base mutation diseases through A to G editing, this brings great application potential for ABEs in disease treatment.

[0004] ABE7.10, as the initial ABEs, contains two TadAs (adenine deaminases): a wild-type TadA and a TadA variant (TadA * ). Subsequently, the researchers obtained a variety of improved ABEs variants, such as ABEmax, miniABEmax, ABE8e, etc., through codon optimization, adding nuclear localization signals or phage-assisted evolution.

[0005] The above-mentioned enhanced ABEs have improved the editing efficiency or increased the editing window to varying degrees. Among them, ABE8e is composed of a single TadA * The domain introduces 8 amino acid mutations (TadA8e), which can show 3 to 11 times higher editing efficiency than ABE7.10. Therefore, ABE8e has been widely used in the treatment of various genetic disease models, including spinal muscular atrophy (SMA), hereditary heart disease, sickle cell anemia, β-thalassemia, cystic fibrosis, etc.

[0006] Although base editors do not induce DNA double-strand breaks (DSBs), the use of deaminases to perform base substitutions at target genomic sites has raised concerns about their genome-wide specificity and potential safety in clinical applications. Both CBEs and ABEs have been shown to induce RNA off-targets at the transcriptome level in cells.

[0007] Given that off-target effects at the genome-wide level are difficult to detect, the inventors’ previous research developed an unbiased genome-wide off-target detection method (GOTI) through two-cell injection of embryos to distinguish spontaneous mutations from off-target mutations, thereby enabling accurate assessment of the specificity of base editors. Using this method, it was found that BE3 produced a large number of genome-wide single nucleotide variations (SNVs) in mouse embryos, while ABE7.10 did not have this effect. However, the specificity at the genome-wide level remains to be explored for optimized ABEs. Summary of the invention

[0008] The purpose of the present invention is to provide the development and application of a high-efficiency, high-fidelity adenine base editor.

[0009] In a first aspect of the present invention, a method for reducing off-target effects of an adenine base editor (ABE) and improving editing accuracy (fidelity) is provided, comprising: using TadA8e as an adenine deaminase, and modifying the amino acid residues of TadA8e; the modification comprises mutating the amino acid residues Y149, S109, N119, N122, D147, R111, I166 or N167 of TadA8e.

[0010] In one or more embodiments, the modification of the amino acid residue includes: mutating the amino acid residue into another amino acid residue (any one of the other 19) other than the amino acid residue.

[0011] In one or more embodiments, the modification of the amino acid residues includes: Y149V, Y149L, Y149M or Y149I for the Y149 site; S109F for the S109 site; N119C, N119Q or N119D for the N119 site; N122F, N122V, N122L or N122A for the N122 site; and D147K for the D147 site.

[0012] In one or more embodiments, the modification of the amino acid residues includes: Y149V and / or S109F (preferably, the mutation is Y149V).

[0013] In one or more embodiments, reducing the off-target effects of the adenine base editor includes reducing the nonspecific binding between the base editor and DNA.

[0014] In one or more embodiments, the mutation of the amino acid position is calculated based on the full length of the TadA8e protein.

[0015] In one or more embodiments, the TadA8e protein includes a protein selected from the following groups: (a1) a protein such as the amino acid sequence of SEQ ID NO: 1; (b1) a protein derived from (a1) which is formed by replacing, deleting or adding one or more (such as 1-20; preferably 1-15; more preferably 1-10, such as 5, 3) amino acid residues of the amino acid sequence of SEQ ID NO: 1, and has the function of the protein of (a1); (c1) a protein derived from (a1) which has more than 80% (preferably more than 85%; more preferably more than 90%; more preferably more than 95%, such as 98%, 99%) homology with the protein sequence defined in (a1) and has the function of the protein of (a1); or (d1) a protein active fragment defined in (a1), or a protein formed by adding a tag sequence, an enzyme cleavage sequence, or a reporter protein to both ends of the protein. The TadA8e also includes the aforementioned biologically active fragments.

[0016] In another aspect of the present invention, a TadA8e variant is provided, which includes mutations of amino acid residues selected from the following group: Y149, S109, N119, N122, D147, R111, I166 or N167; preferably, the mutations include: Y149V, Y149L, Y149M or Y149I for the Y149 site; S109F for the S109 site; N119C, N119Q or N119D for the N119 site; N122F, N122V, N122L or N122A for the N122 site; and / or; D147K for the D147 site; more preferably, the mutations include: Y149V and / or S109F.

[0017] In another aspect of the present invention, an adenine base editor is provided, comprising the TadA8e variant, and an element suitable for adenine editing operably connected thereto; preferably, the adenine base editor comprises: a TadA8e variant, a Cas nuclease operably connected thereto.

[0018] In one or more embodiments, the adenine base editor comprises: an NLS, a TadA8e variant, a Cas nuclease, and an NLS that are operably linked.

[0019] In one or more embodiments, the Cas nuclease includes (but is not limited to): Cas9 nuclease or its homologs, and the homologs include (but are not limited to): SpRY nuclease, SaKKH nuclease, IscB nuclease.

[0020] In one or more embodiments, the NLS is BPNLS.

[0021] In one or more embodiments, a connecting peptide is also included between the TadA8e variant and the Cas nuclease.

[0022] In another aspect of the present invention, an isolated polynucleotide is provided, wherein the polynucleotide encodes the TadA8e variant; or the adenine base editor.

[0023] In another aspect of the present invention, a construct or an expression vector containing the construct is provided, which contains the polynucleotide (encoding the corresponding amino acid sequence).

[0024] In one or more embodiments, the construct includes an expression cassette of the adenine base editor and an operably linked sgRNA expression cassette targeting a target gene.

[0025] In one or more embodiments, the target gene includes (but is not limited to): Hpd, the construct or the expression vector containing the construct is used to prepare a pharmaceutical composition for alleviating or treating hereditary type I tyrosinemia; more preferably, the nucleotide sequence of the sgRNA is as shown in SEQ ID NO:7.

[0026] In one or more embodiments, the construct includes: construct 1: an expression cassette of an sgRNA targeting a target gene, an expression cassette of a TadA8e variant encoding gene and a Cas nuclease N-terminal portion encoding gene; and construct 2: an expression cassette of a Cas nuclease C-terminal portion.

[0027] In one or more embodiments, the construct 1 includes: U6 promoter-sgRNA targeting the target gene-EFS-NLS-TadA8e variant encoding gene-connecting peptide-Cas nucleic acid nickase N-terminal part encoding gene-Intein; preferably, both ends also include ITR; preferably, a connecting peptide is also included between the TadA8e variant encoding gene and the Cas nucleic acid nickase N-terminal part encoding gene; preferably, the NLS is BPNLS.

[0028] In one or more embodiments, the construct 2 includes: EFS-Intein-Cas nucleic acid nickase C-terminal part encoding gene-NLS; preferably, both ends also include ITR; preferably, the TadA8e variant encoding gene and the Cas nucleic acid nickase N-terminal part encoding gene also include a connecting peptide; preferably, the NLS is BPNLS.

[0029] In one or more embodiments, the construct further includes an expression cassette for a reporter protein.

[0030] In another aspect of the present invention, a recombinant cell is provided, which contains the construct or expression vector, or has the polynucleotide (encoding the corresponding amino acid sequence) integrated into its genome.

[0031] In another aspect of the present invention, there is provided a use of the adenine base editor or the construct or an expression vector containing the construct for gene editing or for preparing a reagent for gene editing; wherein the gene editing has reduced off-target effects and improved editing accuracy; preferably, the gene editing is adenine base editor-mediated gene editing; preferably, the gene editing is conversion of A·T to G·C.

[0032] In one or more embodiments, the use of the adenine base editor for gene editing is a cellular level method, including single cells, dual cells or multiple cells.

[0033] In one or more embodiments, the use is applied to isolated cells or cell cultures in vitro.

[0034] In one or more embodiments, the use is not directly intended for disease diagnosis or treatment.

[0035] In another aspect of the present invention, a method for gene editing (including gene editing whose main purpose is the conversion of A·T to G·C) is provided, comprising mediating gene editing with the adenine base editor or the construct or an expression vector containing the construct; preferably, the nucleic acid sequence encoding the adenine base editor is transfected or injected into a receptor to perform gene editing; preferably, the receptor includes: somatic cells or germ cells; preferably, the germ cells include embryonic cells or fertilized eggs.

[0036] In one or more embodiments, the gene editing method is an in vitro method for a non-living organism.

[0037] In one or more embodiments, the subject of the gene editing method will not develop into a living organism.

[0038] In one or more embodiments, the gene editing method is a cellular level method, including single cell, dual cell or multi-cell.

[0039] In one or more embodiments, the gene editing includes gene editing targeting the target gene Hpd; preferably, the reagent for gene editing is used to prepare a pharmaceutical composition for alleviating or treating hereditary tyrosinemia type I; preferably, in the gene editing targeting the target gene Hpd, the nucleotide sequence of sgRNA is as shown in SEQ ID NO:7.

[0040] In another aspect of the present invention, a reagent or kit for gene editing is provided, which contains: the TadA8e variant, or a polynucleotide encoding it; or the adenine base editor, or a polynucleotide encoding it; or the construct or an expression vector containing the construct.

[0041] Other aspects of the present invention will be apparent to those skilled in the art in view of the disclosure herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 , Cas9, BE3, YE1-BE3-FNLS, ABE7.10, ABE7.10 F148A , ABEmax, miniABEmax and ABE8e in 102-sgRNA cells without bias.

[0043] a, Schematic diagram comparing the editing characteristics of Cas9, CBEs, and ABEs in HEK293T cells stably integrating 102 sgRNAs.

[0044] b, Box plots of the targeting efficiency of Cas9, CBEs, and ABEs at 102 sgRNA target sites. Specifically, the efficiencies of indels, C-to-T base substitutions, and A-to-G base substitutions were calculated as the targeting efficiencies of Cas9, CBEs, and ABEs, respectively.

[0045] c, The proportion of indels in different groups.

[0046] d, The proportion of C editing in different groups.

[0047] e, Heat map of the average C-to-T editing efficiency of CBEs and A-to-G editing efficiency of ABEs at sites 1 to 20. Data are from three independent experiments.

[0048] Figure 2 , using the same Tyr sgRNA for Cas9, BE3, YE1-BE3-FNLS, ABE7.10, ABE7.10 F148A , ABEmax, miniABEmax and ABE8e for unbiased genome-wide off-target analysis.

[0049] a, Schematic diagram of the GOTI method for analyzing off-target DNA effects of Cas9 and base editors.

[0050] b, Comparison of off-target SNVs detected. Data are presented as mean ± standard error (n ≥ 3). P values ​​were calculated by two-sided unpaired t-test.

[0051] c, Distribution of SNV mutation types in the Cre, two CBEs, and five ABEs injection groups. Numbers indicate the percentage of a certain type of SNVs among all SNVs.

[0052] d, Sequence motif maps of off-target DNA SNVs from BE3 and ABE8e groups, respectively.

[0053] e, Overlap of off-target SNVs between three BE3-treated embryos and three ABE8e-treated embryos injected with the same Tyr sgRNA.

[0054] f, Overlap between off-target SNVs detected by GOTI and those predicted by Cas-OFFinder and CRISPOR.

[0055] Figure 3 , Cas9, BE3, YE1-BE3-FNLS, ABE7.10, ABE7.10 F148A Analysis of DNA off-target characteristics of ABEmax, miniABEmax and ABE8e.

[0056] a, Targeting efficiency of tdTomato+ and tdTomato- cells in the Tyr region of the treatment groups based on whole genome sequencing (WGS) analysis.

[0057] bc, The proportions of C to T mutations (b) and A to G mutations (c) in the detected off-target DNA SNVs.

[0058] d, Distribution of off-target DNA SNVs on chromosomes in mouse embryos.

[0059] e, Comparison of off-target indels detected. Data are presented as mean ± standard error (n ≥ 3). P values ​​were calculated by two-sided unpaired t-test.

[0060] Figure 4 , unbiased genome-wide off-target analysis of ABE8e using Dmd, Fah, PCSK9, and non-targeting (NT) sgRNAs.

[0061] a, Schematic diagram of sgRNA sites at the Dmd, Fah and PCSK9 gene loci.

[0062] b, Comparison of off-target DNA SNVs detected. Data are presented as mean ± standard error (n ≥ 3). P values ​​were calculated by two-sided unpaired t-test.

[0063] c, Distribution of off-target SNVs mutation types. Numbers indicate the percentage of a certain type of SNVs among all SNVs.

[0064] d, Sequence motif diagram of off-target SNVs.

[0065] e, Overlap of off-target SNVs from different individuals of the same sgRNA.

[0066] f, Overlap between off-target SNVs detected by GOTI and off-target sites predicted by Cas-OFFinder and CRISPOR.

[0067] Figure 5 , Dmd, Fah, PCSK9, and non-targeting (NT) sgRNAs were used to further analyze the off-target properties of ABE8e.

[0068] a, The targeting efficiency of ABE8e at Dmd, Fah, and PCSK9 sites was analyzed based on whole genome sequencing (WGS).

[0069] b, Distribution of off-target DNA SNVs on chromosomes in mouse embryos.

[0070] c, The proportion of A to G mutations in the detected off-target DNA SNVs.

[0071] d, Comparison of off-target indels detected. Data are presented as mean ± standard error (n ≥ 3). P values ​​were calculated by two-sided unpaired t-test.

[0072] Figure 6 , reconstructed the eight mutation sites of TadA8e through saturation mutagenesis.

[0073] a. Schematic diagram of saturation mutation of eight sites in ABE8e.

[0074] b. Editing window of ABE8e and each of its variants.

[0075] c. Peak A-to-G editing efficiency of ABE8e and its variants at the A5 site. The red dashed line is marked at 40% to distinguish between efficiencies comparable to (or higher than) ABE8e and efficiencies lower than ABE8e. Data are presented as mean ± standard error (n = 3).

[0076] d. Line graph of the average A-to-G editing efficiency of 13 variants (S109F, N119Q, N119D, N119C, N122F, N122L, N122V, N122A, D147K, Y149L, Y149I, Y149M and Y149V) at sites 1-20.

[0077] e. Bystander A and undesired C editing for 13 variants. A3-A7 is defined as within the target window. A2 / A5 and A8 / A5 represent the ratio of undesired A2 and A8 editing to the desired peak A5 editing; C5 / A5 represents the ratio of undesired C5 editing to the desired A5 editing. Data are presented as mean ± standard error (n=3).

[0078] f. Line graph of the average A-to-G editing efficiency at sites A1-A14 of six double mutants (S109F / N119Q, S109F / D147K, S109F / Y149V, N119Q / D147K, N119Q / Y149V and D147K / Y149V).

[0079] g. Heat map showing the average A-to-G editing efficiency at sites A1-A14 for the three selected variants in this study (ABE8eS109F, ABE8eY149V, ABE8eS109F / Y149V) and previously reported high-fidelity variants (ABE8eV106W, ABE8eV82G, ABE8eK20A / R21A, ABE8eF148A, ABE8eN108Q, or ABE9).

[0080] Figure 7 、ABE8e S109F 、ABE8e Y149V and ABE8e S109F / Y149V Genome-wide and transcriptome-wide off-target analysis.

[0081] a.ABE8e S109F 、ABE8e Y149V and ABE8e S109F / Y149V Comparison of off-target DNA SNVs detected in the groups. Data are presented as mean ± standard error (n = 3). P values ​​were calculated by two-sided unpaired t-test.

[0082] b.ABE8e S109F 、ABE8e Y149V and ABE8e S109F / Y149V Distribution of DNA mutation types in the group.

[0083] c. From ABE8e S109F 、ABE8e Y149V and ABE8e S109F / Y149V Sequence motif diagram of off-target DNA SNVs in the group.

[0084] d. ABE8e analyzed by RNA-seq S109F 、ABE8e Y149V and ABE8e S109F / Y149VThe number of off-target RNA SNVs in the group. Data are presented as mean ± standard error (n = 3). P values ​​were calculated by two-sided unpaired t-test.

[0085] e.GFP, ABE8e, ABE8e S109F 、ABE8e Y149V and ABE8e S109F / Y149V Distribution of RNA mutation types in the group.

[0086] f. Motif map of RNA SNVs.

[0087] Figure 8 、TadA8e Y149V Compatibility with different CRISPR systems.

[0088] a. ABE8e and ABE8e based on SpRY Y149V Frequency of A to G editing at 10 sites with NNN PAM. In the heatmap, the editing efficiencies shown represent the means of three biologically independent replicates.

[0089] b. ABE8e and ABE8e based on SpRY Y149V Average A to G editing efficiency in the A2-A14 region at 10 sites.

[0090] c. ABE8e and ABE8e based on SpRY Y149V Average C to T / G / A editing efficiency at 7 target sites. Data are presented as mean ± SE (n=3).

[0091] d. SaKKH-based ABE8e and ABE8e Y149V Heat map of A to G editing frequencies at six sites with NNNRRT PAMs.

[0092] e. SaKKH-based ABE8e and ABE8e Y149V Average A to G editing efficiency in the A2-A22 region at six sites.

[0093] f. IscB-based ABE8e and ABE8e Y149V Heat map of A to G editing efficiency at five NWR RNA PAM sites.

[0094] g. IscB-based ABE8e and ABE8e Y149V Average A to G editing efficiency in the A1-A16 region at five sites.

[0095] Fig. 9 、ABE8e Y149V Treatment of mice with hereditary tyrosinemia type I.

[0096] a.ABE8e Y149V Schematic diagram of mediated knockout of mouse Hpd gene. Hpd sgRNA was designed to target the start codon (ATG) of Hpd gene, positioning the targeting A at position 6. The mutation from A to G can induce the knockout of Hpd gene.

[0097] b.ABE8e Y149V Schematic diagram of the dual AAV system for in vivo delivery of Fah and Hpd sgRNA. - / - Mice were injected with dual AAVs and treated with nitisinone for 7 days followed by drug withdrawal.

[0098] c.ABE8e Y149V -Hpd sgRNA and saline-treated Fah - / - Mouse body weight ratio. Data are expressed as mean ± standard error (n = 3).

[0099] d. Fah on day 30 after injection - / - Deep sequencing analysis of the Hpd genomic region in mouse liver DNA. Data are presented as mean ± standard error (n = 5).

[0100] e. On the 30th day after injection, the Fah - / - Fah mice treated with dual AAVs without nitisinone - / - Mouse livers were subjected to immunohistochemistry (IHC) staining with anti-HPD antibody.

[0101] fh. Fah who did not receive injection but took Nitisinone - / - Mice, Fah injected with saline but not receiving nitisinone - / - mice, and Fah mice injected with dual AAVs but without nitisinone - / - Mice were subjected to serum level determination, including aspartate aminotransferase (AST), alanine aminotransferase (ALT) and total bilirubin. Data are presented as mean ± standard error (n = 5). P values ​​were calculated by two-sided unpaired t-test. DETAILED DESCRIPTION

[0102] In the present invention, the activity of several adenine base editors (ABE) for a large number of sgRNA target sites was compared, and it was found that ABE8e exhibited the highest editing efficiency and a wider editing window; further analysis showed that ABE8e introduced serious sgRNA-independent DNA off-target effects. To this end, the inventors conducted an in-depth analysis of ABE8e, revealing some sites related to off-target effects and editing accuracy, mutated these sites, and optimized a series of TadA8e variants, which significantly reduced the off-target effects of adenine base editors and improved editing accuracy (fidelity).

[0103] As used in the present invention, "mutation" refers to the substitution of a residue in a sequence (e.g., a nucleic acid or amino acid sequence) with another residue or the change of one or more residues in a sequence to another residue, or the occurrence of a deletion or insertion. In the editor of the present invention, it is hoped to enhance the targeted editing performance, produce as many "A·T to G·C" conversions as possible, and reduce the conversion of other types of bases or other effects; the enhanced targeted editing performance includes: reduced off-target effects, improved editing accuracy, improved editing efficiency, improved targeting specificity, etc.

[0104] As used herein, unless otherwise specified, the "TadA8e mutant", "TadA8e variant" and "mutant TadA8e" are used interchangeably and refer to the amino acid sequence corresponding to TadA8e, wherein the mutation is selected from the following positions or a combination thereof: Y149, S109, N119, N122, D147.

[0105] If it is necessary to indicate TadA8e before mutation, it may be an enzyme with an amino acid sequence such as SEQ ID NO: 1. Unless otherwise specified, the mutation site of the mutant in the present invention is based on the sequence shown in SEQ ID NO: 1.

[0106] In the present invention, unless otherwise specified, the mutant is identified by "amino acid replaced by the original amino acid position" to indicate the mutated amino acid in the mutant, such as Y149V, indicating that the amino acid at position 149 is replaced by V from Y of the starting enzyme.

[0107] As used herein, "operably linked" or "operably connected to" refers to a situation where certain parts of a linear DNA sequence can regulate or control the activity of other parts of the same linear DNA sequence. For example, if a promoter controls the transcription of a sequence, it is operably linked to a coding sequence. "Operationally linked (connected)" means that the connection of each element is in a certain order, such as from amino acid to carboxyl terminus.

[0108] As used in the present invention, the "element" refers to a series of functional nucleic acid / protein sequences useful for the expression of proteins, and the "element" is systematically constructed to form an expression construct. The sequences of the "element" can be those provided in the present invention, and also include their variants, as long as these variants substantially retain the function of the "element", which are obtained by inserting or deleting some bases (such as 1-50bp; preferably 1-30bp, more preferably 1-20bp, more preferably 1-10bp), or performing random or site-directed mutagenesis, etc.

[0109] As used in the present invention, the "construct" refers to a single-stranded or double-stranded DNA molecule that has been artificially intervened to contain DNA fragments combined and arranged according to a sequence that does not exist in nature. The "construct" includes "plasmid", "in vitro transcription product" or viral vector, etc.; or, the "construct" is contained in an expression vector as a part of the expression vector.

[0110] As used in the present invention, the "sgRNA" is a "single-guide RNA (sgRNA)", which is designed based on the "target site on the target gene", and the sequence it contains is sufficient to cooperate with the endonuclease Cas to guide the Cas-mediated DNA double-strand break at the target site. In the present invention, the "sgRNA" includes sgRNA in RNA form (such as mRNA form), and also includes a sequence in DNA form corresponding to the sgRNA sequence or a construct containing the sequence, as long as they can be processed or converted into active "sgRNA" in the cell.

[0111] In the present invention, the "target gene" may also be referred to as "purpose gene", which refers to a gene of interest that is useful for observing cell changes, regulating cell performance or improving cell-related diseases after being introduced into cells.

[0112] As used in the present invention, the "mammal" is not particularly limited, as long as its cells have a genome in the general sense and the gene editing system is active in its cells. For example, the animal can be an animal of the class Mammalia, including humans, non-human primates (monkeys, gorillas), domestic and farm animals (e.g., pigs, sheep, cattle), mice (mice), and rodents (e.g., mice, rats, rabbits), etc.

[0113] As used in the present invention, the "cell" includes but is not limited to: somatic (tissue) cells, induced pluripotent stem cells, and reproductive cells (such as fertilized egg cells and oocytes).

[0114] As used in the present invention, the "off-target effect" refers to the failure of the modification of a specific position in the genome to achieve the pre-set target, the modification is offset or the modification fails to occur. The reasons for the "off-target effect" include but are not limited to: inaccurate binding to the target, inaccurate cutting operation after recognition sequence, insufficient accuracy of editing of the cutting site, etc.

[0115] Although genome editing technology has been widely used in recent years, it still has problems such as off-target effects. How to use this technology to correct the cell genome and change the disease state is still a difficult task. For most diseases in the existing technology, no effective treatment methods and therapeutic drugs based on gene editing technology have been found.

[0116] In the present invention, genome-wide off-target analysis by two-cell embryo injection (GOTI) showed that ABE8e introduced serious sgRNA-independent DNA off-target effects. To address this problem, the inventors conducted in-depth research and analysis on TadA8e, generating 152 single mutants and 6 double mutants by saturation mutagenesis of eight bases.

[0117] Based on the new discovery of the inventors, a method for reducing the off-target effect of an adenine base editor (ABE) and improving editing accuracy (fidelity) is provided, comprising: using TadA8e as an adenine deaminase, and modifying the amino acid residues of TadA8e; the modification comprises mutating the amino acid residues Y149, S109, N119, N122, D147, R111, I166 or N167 of TadA8e.

[0118] In the mutant, ABE8e Y149V It was identified as the optimal variant due to its high efficiency comparable to ABE8e and no off-target effects detected in both DNA and RNA. Y149V Compatible with different Cas homologs, thus expanding the scope of editing applications. In addition, ABE8e was used for double adeno-associated virus (AAV)-mediated Y149V The present invention highlights that the ABE8e mutant, especially ABE8e Y149VTherapeutic implications in ABE-mediated gene therapy.

[0119] The present invention also includes fragments, derivatives and analogs of the TadA8e mutant. As used herein, the terms "fragment", "derivative" and "analog" refer to proteins that substantially retain the same biological function or enzymatic activity as the TadA8e mutant of the present invention. The fragments, derivatives or analogs of TadA8e may be (i) proteins in which one or more conservative or non-conservative amino acid residues (preferably conservative amino acid residues) are substituted, and such substituted amino acid residues may or may not be encoded by the genetic code, or (ii) proteins having substitution groups in one or more (such as 1-20, preferably 1-10, and even more preferably 1-8, 1-5, 1-3, or 1-2) amino acid residues, or (iii) proteins formed by the fusion of an additional amino acid sequence to this protein sequence (such as a leader sequence or secretory sequence or a sequence or proprotein sequence used to purify this protein, or a fusion protein). According to the definition herein, these fragments, derivatives and analogs belong to the scope known to those skilled in the art. However, the condition that needs to be met is that the amino acid sequence of the TadA8e mutant and its fragments, derivatives and analogs must contain at least one mutation specifically mentioned above in the present invention, namely mutations occurring at the following positions: Y149, S109, N119, N122, D147.

[0120] For example, in the art, when amino acids with similar or similar properties are substituted, the function of the protein is usually not changed. For another example, adding or deleting one or more amino acids at the C-terminus and / or N-terminus usually does not change the function of the protein. The term also includes enzymatically active fragments and enzymatically active derivatives of TadA8e mutants. However, in these variant forms, there must be at least one mutation in the key position described above in the present invention.

[0121] In the adenine base editor of the present invention, each element can be a recombinant protein or a synthetic protein, preferably a recombinant protein.

[0122] The present invention also includes variants formed by making changes to individual elements while retaining their functions, that is, function-conservative variants (including fragments, derivatives, and analogs, etc.).

[0123] The present invention also provides a polynucleotide sequence encoding the selected optimized or modified enzyme variant or its conservative variant protein of the present invention.

[0124] The polynucleotide encoding the mature protein of the mutant includes: a coding sequence encoding only the mature protein; a coding sequence of the mature protein and various additional coding sequences; a coding sequence of the mature protein (and optional additional coding sequences) and non-coding sequences. "Polynucleotide encoding a protein" may include a polynucleotide encoding the protein, or may include additional coding and / or non-coding sequences.

[0125] The full-length nucleotide sequence of the enzyme selected for optimization or modification of the present invention or its fragment can usually be obtained by PCR amplification, recombination or artificial synthesis. For PCR amplification, primers can be designed based on the relevant nucleotide sequence disclosed in the present invention, especially the open reading frame sequence, and a commercially available cDNA library or a cDNA library prepared by conventional methods known to those skilled in the art is used as a template to amplify the relevant sequence. When the sequence is long, it is often necessary to perform two or more PCR amplifications, and then splice the fragments amplified in each time together in the correct order.

[0126] Once the relevant sequence is obtained, it can be obtained in large quantities by recombinant methods. This is usually done by cloning it into a vector, then transferring it into cells, and then isolating the relevant sequence from the propagated host cells by conventional methods.

[0127] The present invention also relates to a vector comprising the polynucleotide of the present invention, a host cell produced by genetic engineering using the vector of the present invention or a selected, optimized or modified enzyme coding sequence, and a method for producing the protein of the present invention by recombinant technology.

[0128] The vector containing the above-mentioned appropriate DNA sequence and an appropriate promoter or control sequence can be used to transform appropriate host cells or recipient cells.

[0129] The present invention also provides a method for gene editing, including gene editing mediated by the adenine base editor of the present invention. In addition to using the adenine base editor of the present invention for gene editing, other gene editing reagents can be used as known in the art. In the embodiments of the present invention, preferred constructs and implementation methods are provided.

[0130] In the present invention, there is no particular limitation on the objects of gene editing, and they can be somatic cells or germ cells, animal cells or human cells.

[0131] In a preferred embodiment of the present invention, a construct that can be used for targeted treatment of hereditary tyrosinemia type I is provided, which includes an expression cassette of an adenine base editor carrying the TadA8e mutant of the present invention, and an operably linked sgRNA expression cassette targeting a target gene; the target gene is Hpd.

[0132] In conjunction with the above-mentioned adenine base editor, the present invention also provides a preferred sgRNA for Hpd. In a preferred embodiment, the sgRNA is a sgRNA of the sequence shown in SEQ ID NO:7.

[0133] The sgRNA can be introduced into the cell separately or together with the adenine base editor.

[0134] The sgRNA and the adenine base editor can be included in a pharmaceutical composition.

[0135] The pharmaceutical composition may also contain a pharmaceutically acceptable carrier. Such carriers include (but are not limited to): saline, buffer, glucose, water, glycerol, ethanol, and combinations thereof. Usually, the pharmaceutical preparation should match the mode of administration. The pharmaceutical composition of the present invention can be prepared in the form of an injection, for example, by conventional methods using physiological saline or an aqueous solution containing glucose and other adjuvants. The pharmaceutical composition is preferably manufactured under sterile conditions. The dosage of the active ingredient is a therapeutically effective amount.

[0136] The present invention is further described below in conjunction with specific examples. It should be understood that these examples are only used to illustrate the present invention and are not intended to limit the scope of the present invention. The experimental methods in the following examples where specific conditions are not specified are usually carried out according to conventional conditions such as those described in J. Sambrook et al., Molecular Cloning Experiment Guide, Science Press, or according to the conditions recommended by the manufacturer.

[0137] Materials and methods

[0138] 1. Vector construction

[0139] Base Editors:

[0140] (1) Cytosine base editor 3 (BE3): rAPOBEC1-nCas9-UGI (Cas9 is spCas9).

[0141] (2) BE3-FNLS: BE3 of (1), in which the N-terminus of the rAPOBEC1 sequence is connected to the Flag tag and the NLS nuclear localization sequence (sequence: PKKKRKV), and there is also an NLS at the C-terminus of the base editor.

[0142] (3) YE1-BE3-FNLS: For BE3-FNLS, the 90th position of the rAPOBEC1 sequence was mutated from W to Y (W90Y), and the 126th position was mutated from R to E (R126E).

[0143] (4)ABE7.10 F148A: ABE7.10 editor, in which the 148th F in the TadA and TadA* sequences were mutated to A.

[0144] (5)ABE8e: BPNLS, TadA8e, linker, nCas9, BPNLS, polyA.

[0145] (6)ABE8e S109F : BPNLS, TadA8e (position 109 mutated from S to F), linker, nCas9, BPNLS, polyA.

[0146] (7)ABE8e Y149V : BPNLS, TadA8e (position 149 mutated from Y to V), linker, nCas9, BPNLS, polyA.

[0147] (8)ABE8e-nSpRY: BPNLS, TadA8e, linker, nSpRY, BPNLS, polyA.

[0148] (9)ABE8e Y149V -nSpRY: BPNLS, TadA8e (position 149 mutated from Y to V), linker, nSpRY, BPNLS, poly.

[0149] (10)ABE8e-nSaKKH: BPNLS, TadA8e, linker, nSaKKH, BPNLS, polyA.

[0150] (11)ABE8e Y149V -nSaKKH: BPNLS, TadA8e (position 149 mutated from Y to V), linker, nSaKKH, BPNLS, poly.

[0151] (12)ABE8e-enIscB: BPNLS, TadA8e, linker, enIscB, BPNLS, polyA.

[0152] (13)ABE8e Y149V -enIscB: BPNLS, TadA8e (position 149 mutated from Y to V), linker, enIscB, BPNLS, polyA.

[0153] Among them, the linker sequence is: SGGSSGGSSGSETPGTSESATPESSGGSSGGS.

[0154] The amino acid sequence of adenine deaminase TadA8e is as follows (SEQ ID NO: 1):

[0155] MSEVEFSHEYWMRHALTLAKRARDEREVPVGAVLVLNNRVIGEGWNRAIGLHDPTA HAEIMALRQGGLVMQNYRLIDATLYVTFEPCVMCAGAMIHSRIGRVVFGVRNS 109 KR 111 GAAGSLMN 119 VLN 122 YPGMNHRVEITEGILADECAALLCD 147 FY 149 RMPRQVFNAQK KAQSSI 166 N 167

[0156] The editor expression cassette (including Cas9, CBE and ABE) was cloned into the pCMV-editor-backbone vector, and the sgRNA was cloned into the sgRNA vector backbone. Gene templates were obtained from Addgene or synthesized in GENEWIZ. PCR amplification was performed using KOD-Plus-Neo high-fidelity DNA polymerase (Toyobo, KOD-401). Recombinant plasmids or different ABE8e variants were constructed using NEBuilder HiFi DNA Assembly Master Mix (New England BioLabs). sgRNA targeting oligonucleotides were synthesized, annealed and ligated to BbsI or BsaI sites to generate sgRNA expression vectors. The protoregions and PAM sequences of all targets are listed in Table 1.

[0157] Table 1. Protospacer and PAM sequences

[0158]

[0159] 2. Cell culture, transfection, FACS and genomic DNA extraction

[0160] HEK293T and 102-sgRNA cells were cultured in modified Dulbecco's medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (FBS; Biological Industries) and 1% penicillin / streptomycin (Beyotime) at 37°C in a 5% CO2 incubator. To compare editing efficiencies in parallel, the editor plasmid was transfected into 102-sgRNA cells alone or co-transfected with the sgRNA expression vector into HEK293T cells using PEI (Polyscience). 48 hours after transfection, double-positive cells expressing GFP and mCherry were isolated by flow cytometry sorting (FACS, BD FACSAria III). Approximately 500,000 102-sgRNA cells were sorted, and genomic DNA (gDNA) was extracted using the TIANamp Genomic Kit (Tiangen). In addition, approximately 20,000 HEK293T cells were sorted and gDNA was extracted using the One-Step Mouse Genotyping Kit (Vazyme).

[0161] 3. Deep sequencing of target sites

[0162] Takara Ex Taq polymerase (Takara) was used to perform nested PCR (Table 2) to amplify the genomic sequence of the target site. Barcodes were added in the second round of PCR to distinguish samples. PCR products were purified using a universal DNA purification kit (TIANGEN) and 150-bp paired-end sequencing was performed on the Illumina NovaSeq 6000 platform (Genewiz Co. Ltd). The raw data of deep sequencing were multiplexed using fastq-multx, and the read sequences of each sample were aligned to the reference target sequence using CRISPResso2 (v2.0.32). The in-situ editing efficiency of each target was calculated using an in-house script (v5.26.2).

[0163] Table 2. Primers for target regions

[0164]

[0165]

[0166]

[0167] 4. Animal husbandry

[0168] Heterozygous Ai9 (B6.Cg-Gt(ROSA)26Sortm9(CAG-td-Tomato)Hze / J; JAX 007909) male mice were mated with 4-week-old female C57BL / 6 mice to collect embryos. ICR female mice were used as recipients. Fah was used in in vivo editing experiments - / - Mice. The use and care of animals complied with life science ethics.

[0169] 5. In vitro transcription of mRNA and sgRNA

[0170] The plasmid was amplified by PCR using primers IVT F and R (Table 3) to add a T7 promoter to the N-terminus of the coding region. The purified PCR product was used as a template for in vitro transcription (IVT) using the mMESSAGE mMACHINE T7 ULTRA kit (Life Technologies). For sgRNA IVT, the T7 promoter was added by PCR amplification of pX330, and the purified PCR product was used as a template for IVT using the MEGA shortscript T7 kit (Life Technologies). Editor mRNA and sgRNA were purified using the MEGA clear kit (Life Technologies).

[0171] Table 3. In vitro transcription primers

[0172]

[0173] 6. Two-cell embryo injection, embryo transfer and FACS

[0174] Four-week-old superovulated C57BL / 6 female mice were mated with homozygous Ai9 male mice, and fertilized embryos were obtained from the oviduct 24 hours after hCG injection. For two-cell embryo injection, a mixture of Cas9 or base editor mRNA (50 ng / μl), sgRNA (50 ng / μl), and Cre mRNA (2 ng / μl) was injected into one blastomere of two-cell embryos 48 hours after hCG injection. Injections were performed using a FemtoJet microinjector (Eppendorf) in a droplet of M2 medium containing 5 μg / ml cytochalasin B (CB) with a constant flow setting. After injection, embryos were cultured in KSOM medium containing amino acids for 2 hours and then transferred to the oviduct of pseudopregnant ICR female mice at 0.5 dpc. At embryonic day 14.5 (E14.5), the embryonic tissue was cut into small pieces and digested with 5 mL 0.05% trypsin-EDTA (Gibco) at 37°C for 30 minutes. Digestion was terminated by adding 5 mL of DMEM medium containing 10% FBS. Subsequently, the fetal tissue was homogenized with a 1 mL pipette tip and the cell suspension was centrifuged at 200 g for 6 minutes. The resulting pellet was resuspended in 2 mL of DMEM containing 10% FBS. Finally, the cell suspension was filtered through a 40 μm cell strainer and sorted by flow cytometry to separate tdTomato+ and tdTomato- cells.

[0175] 7. Whole genome sequencing (WGS) and data analysis

[0176] tdTomato was extracted using DNeasy blood and tissue kit (Qiagen) + and tdTomato - gDNA of cells. WGS was performed at an average coverage of 50× using the BGIDNBSEQ-T7 platform. Qualified sequencing reads were aligned to the reference genome (GRCm39) using BWA (v0.7.17). The aligned reads were sorted and duplicates of the mapped BAM files were marked using the Picard tool (v2.25.7). Subsequently, whole-genome SNVs were called using three algorithms: Mutect2 (v3.5), Lofreq (v2.1.2), and Strelka (v2.7.1), respectively, with default parameters. In parallel, whole-genome indels were detected using Mutect2 (v3.5), Scalpel (v0.5.3), and Strelka (v2.7.1), with default parameters. The overlapping parts of the SNVs or indels of the three algorithms were considered to be true variants.

[0177] 8. RNA library preparation, sequencing and data analysis

[0178] HEK293T cells were transfected with base editors and sgRNA plasmids, or with GFP and mCherry plasmids as controls. After 48 hours, approximately 500,000 cells (top 5% GFP / mCherry) were collected and RNA was extracted with Trizol (Ambion). For library construction, mRNA was first fragmented and converted to cDNA with random primers or oligo (dT) primers. Subsequently, the adapter was connected to the 5' and 3' ends of the cDNA fragments, and the correctly connected fragments were enriched and PCR amplified. Sequencing was performed on the Illumina Novasek 6000 platform. For RNA-seq data analysis, initial quality control used FastQC (v0.11.3). Subsequently, STAR (v2.7.1) aligned the processed reads to the GRCh38 reference genome in 2-pass mode with default parameters. After alignment, the Picard tool (v2.25.5) was used to sort the aligned reads and identify duplicates in the BAM file. Variants were called using HaplotypeCaller of GATK (v4.2.0.0).

[0179] 9. AAV vector cloning, production and in vivo injection

[0180] By using U6-Hpd sgRNA-N-TadA8e Y149V -nCas9(2-573)-intein and C-intein-nCas9(574-1368) sequences were cloned between the two ITRs of AAV to construct AAV vectors. These vectors, together with the helper plasmid and AAV8 packaging plasmid, were transfected into HEK293T cells using PEI. Viral particles were collected from cell culture media and cells 3-5 days after transfection. Subsequently, the collected particles were purified and concentrated.

[0181] Fah aged 6 to 8 weeks - / - 5×10 per kg body weight were injected intravenously into mice 11 (vg / kg) of dual AAV mixture, the control group of 6 to 8 weeks old Fah - / - Mice were injected with 200 μl of saline only.

[0182] 10. Mouse Body Weight, Blood Collection, and Serum Analysis

[0183] The injected mice were provided with 10 mg / L of nitisinone (Sigma-Aldrich) as drinking water for the first 7 days, after which nitisinone feeding was stopped. The body weight of the mice was recorded every other day. From the control mice that stopped nitisinone treatment, euthanasia was performed when the body weight loss exceeded 20%. Blood was collected retro-orbitally, and after clotting for 2 hours at room temperature, serum was separated by centrifugation at 3000g for 15 minutes at 4°C. Aspartate aminotransferase (AST), alanine aminotransferase (ALT) and total bilirubin were determined using fresh serum samples.

[0184] 11. Immunohistochemical staining

[0185] Liver tissue was fixed with 4% paraformaldehyde (PFA) at 4°C for 2 hours, then dehydrated with ethanol and embedded in paraffin. Paraffin sections were 5 μm thick, dewaxed with xylene, and rehydrated with ethanol and distilled water. Anti-HPD antibody (ab133515, Abcam) was used as the primary antibody. The sections were sealed with glass slides and photographed with a bright field microscope system (Leica Microsystems).

[0186] Example 1: ABE8e shows higher editing efficiency and wider editing window than other ABE tools

[0187] In order to identify the most suitable ABE gene editor for further improvement, five ABEs (ABE7.10, ABE7.10 F148A , ABEmax, miniABEmax, ABE8e) were compared in parallel, and wild-type Cas9 and CBE (BE3, YE1-BE3-FNLS) were used as controls. To ensure a comprehensive and unbiased evaluation, the inventors adopted an sgRNA target library detection strategy. In brief, a plasmid library containing 102 sgRNAs and their corresponding target sequences was constructed and integrated into the genome of HEK293T cells by lentiviral infection (hereinafter referred to as 102-sgRNA cells). These sgRNAs contain at least one adenine within 1-20 nucleotides from the end of the protospacer sequence adjacent motif (PAM). Eight different editors were transfected into 102-sgRNA cells, and the editing efficiencies of these ABEs, CBEs, and Cas9s were calculated based on the frequencies of A-to-G base substitutions, C-to-T base substitutions, and indels at the target site ( Figure 1 a).

[0188] Deep sequencing analysis of 102 targets in each editor treatment group showed that the on-target editing efficiency of ABE8e was significantly higher than that of other ABEs (average 35.9%, all other ABEs were ≤15.2%) ( Figure 1b). In comparison, Cas9 showed an average editing efficiency of 36.4%, while the average C-to-T editing efficiencies of the BE3 and YE1-BE3-FNLS groups were 13.6% and 18.8%, respectively.

[0189] However, evaluation of the purity of the editing products showed that ABE8e also introduced relatively high levels of indels and cytosine edits ( Figure 1 c and 1d). In addition, ABE8e showed the widest editing window, which was significantly wider than that of the other four ABEs and two CBEs ( Figure 1 e).

[0190] These results indicate that ABE8e provides high gene editing efficiency and a wide editing window among these common base editors.

[0191] Example 2: ABE8e introduces significantly more off-target effects than other ABEs across the genome

[0192] To comprehensively evaluate the genome-wide off-target effects mediated by these five ABEs, GOTI tests were performed using sgRNAs targeting the tyrosinase (Tyr) gene (Tables 1-2). F148A mRNAs of ABEmax, miniABEmax, ABE8e or Cas9, BE3, YE1-BE3-FNLS control were injected into one blastomere of Ai9(CAG-LoxP-Stop-LoxP-tdTomato) mouse two-cell embryos together with Tyr sgRNA and mRNA encoding Cre. Edited cells (tdTomato + ) and unedited cells (tdTomato - )( Figure 2 a). After verification of targeting efficiency by Sanger sequencing, whole-genome sequencing (WGS) was performed on edited and unedited cells (50× coverage) from all eight editing treatment groups (at least three embryos per group). Single nucleotide variants (SNVs) and insertion / deletion variants (indels) were then identified in edited cells using three independent algorithms (Mutect2, Lofreq, and Strelka), with unedited cells from the same embryo as a reference.

[0193] WGS analysis further confirmed the targeting efficiency of Cas9, CBE, and ABE editing tools ( Figure 3a). Off-target effect detection found that there was a relatively high number of SNVs in cells edited with BE3 (an average of 225 per embryo). Unexpectedly, the results also showed that the number of off-target mutations caused by ABE8e using the same Tyr sgRNA was significantly higher than that of BE3 (an average of 371 per embryo), which was more than 28 times that of the Cre-only control group (i.e., 13 SNVs per embryo). In contrast, the average number of SNVs for the other four ABEs, Cas9, and YE1-BE3-FNLS was between 12 and 21 per embryo, which was not significantly different from the Cre-only control embryos ( Figure 2 b).

[0194] The distribution analysis of base conversion types in each editor group showed that the off-target editing of the BE3 group was significantly biased towards C-to-T and G-to-A (accounting for 95% of the total off-target), while the ABE8e group was biased towards A-to-G and T-to-C base conversions (accounting for 94% of the total off-target) ( Figure 2 In addition, the sequence logos showed that BE3 usually introduced off-targets in the context of TC sequences, while ABE8e tended to introduce off-targets at TA sites ( Figure 2 d). These preferences are identical to those observed for the cytosine deaminase APOBEC1 and the adenine deaminase TadA8e, respectively. Furthermore, almost no off-target sites were shared between ABE8e-treated embryos ( Figure 2 e), nor did they overlap with predicted off-target mutations ( Figure 2 f), which is similar to the off-target characteristics of BE3. These new mutations are randomly distributed on the chromosome ( Figure 3 d). This finding indicates that ABE8e-mediated off-target SNVs are sgRNA-independent, suggesting that they are caused by TadA8e. The inventors also noted that the number of indels in all eight experimental groups was slightly increased relative to the Cre control group, although the average number of indels per embryo did not exceed 13 ( Figure 3 e). These results indicate that the off-target frequency introduced by ABE8e in mouse embryos is significantly higher than that of other commonly used base editors.

[0195] Since ABEs are frequently used to introduce or reverse point mutations in disease-associated genes, mouse disease models can be created or studied. To further validate the sgRNA-independent off-target effects of ABE8e, GOTI testing was performed using three different sgRNAs targeting the disease-associated genes Dmd, Fah, and PCSK9. In addition, a target site that does not exist in the human or mouse genome, called a non-targeting (NT) sgRNA ( Figure 4a). The inventors observed that ABE8e-mediated targeted editing efficiencies of sgRNAs for Dmd (average 85.8%), Fah (average 73.6%), and PCSK9 (average 91.3%) were high ( Figure 5 a).

[0196] Consistent with the off-target analysis results for Tyr sgRNA, the average number of SNVs caused by ABE8e on Dmd, Fah, PCSK9, and NT sgRNAs was 279, 343, 387, and 313, respectively, which was 21 to 30 times that of the Cre-only control group ( Figure 4 b and 5b), where A-to-G and T-to-C substitutions accounted for 91%-95% of the off-target types ( Figure 4 c and 5c).

[0197] Notably, despite reports that ABE can catalyze cytosine conversion, the off-target C-to-T / G / A cytosine deamination activity detected by the inventors in these sgRNAs was not significant. All off-target sites of the four different sgRNAs showed a consistent preference for converting A to G within the TA motif ( Figure 4 d). In addition, no significant overlap of off-target SNVs was detected between replicate experiments ( Figure 4 e), these SNVs also did not overlap with predicted off-target mutations ( Figure 4 f). Similarly, the number of indels of ABE8e and these four sgRNAs increased ( Figure 5 d).

[0198] These results demonstrate that ABE8e introduces off-target effects genome-wide in an sgRNA-independent manner.

[0199] Example 3, mutation analysis of TadA8e

[0200] In this example, 102-sgRNA cells were used for testing to analyze the editing properties of different mutants at 102 sgRNAs sites.

[0201] Furthermore, the inventors hope to improve the accuracy of ABE8e by narrowing its editing window. Analysis and comparison of the TadA deaminase components of ABE7.10, ABEmax and miniABEmax showed that TadA8e contained eight mutation sites, S109, R111, N119, N122, D147, Y149, I166 and N167, relative to these three ABEs ( Figure 6 a). In combination with the inventors Figure 2Based on the findings in the previous article that these three ABEs do not cause significant genome-wide off-target SNVs, the inventors hypothesized that the eight mutations in TadA8e are responsible for its increased off-target activity. In order to reduce off-target editing while maintaining its high targeting efficiency, a saturation mutagenesis strategy targeting these eight sites in ABE8e was designed.

[0202] 152 ABE8e single variants were generated by saturation mutagenesis of eight bases. Each variant was then transfected into 102-sgRNA cells and subjected to targeted deep sequencing analysis. To ensure uniform comparison between variants, the editing window was defined as the target sequence position where the average editing efficiency was greater than 30% of the average peak editing efficiency. Figure 6 As shown in Fig. 6b and 6c, with PAM positions 21-23, the base editing activity window of ABE8e is from A2 to A10, and the highest editing efficiency occurs at A5 (average 44.0%). Comparison of the activities of the eight site mutants showed that the mutation at the R111 position significantly reduced the efficiency, while the mutations at the I166 and N167 positions had no significant effect on the size of the editing window.

[0203] It is worth noting that 16 of the 19 mutations at the S109 position resulted in a narrower editing window, of which S109F had the narrowest editing window (A3-A7) and the highest editing efficiency (A5 average activity of 42.3%). All mutations at the N119 position narrowed the editing window by 1 to 4 nucleotides, of which the N119Q (A3-A9), N119D (A3-A10) and N119C (A3-A10) variants had higher editing efficiencies than ABE8e (average 47.8%, 46.8% and 48.6%, respectively). At the N122 position, four mutants (N122F, N122L, N122V and N122A) showed relatively high editing efficiencies (48.1%, 46.6%, 47.9% and 41.0%, respectively), and the editing window was one base narrower than ABE8e. At the D147 position, the editing window of the D147K mutant was slightly narrowed (A2-A9), and its editing efficiency was significantly higher than that of ABE8e, reaching 53.1%. Most Y149 mutants had a strict editing window (A3-A7), some of which showed higher editing efficiency, including Y149L (average 45.1%), Y149I (40.6%), Y149M (45.9%), and Y149V (42.4%).

[0204] These 13 variants were further compared for A-to-G editing, bystander A editing, and non-target cytosine conversion at different protospacer positions (S109F, N119Q, N119D, N119C, N122F, N122L, N122V, N122A, D147K, Y149L, Y149I, Y149M, and Y149V; Figure 4 d and 4e) show that Y149V has the narrowest editing window and the lowest bystander A editing and C editing, indicating that it is the best candidate among the single mutants. The inventors also noted that the S109F and N119Q mutations at the other two positions ranked second and third, respectively. At the same time, D147K showed the highest efficiency at positions A3 to A8 and had less C editing relative to ABE8e, so this mutation was also included in the subsequent combined mutation test.

[0205] To examine whether combining these effective mutations can further narrow the editing window and / or improve A-to-G editing efficiency, the inventors combined S109F, N119Q, D147K and Y149V in pairs to generate six double mutants. Among them, the editing window of S109F / Y149V is relatively narrow (A3-A6), and the editing efficiency at position A5 is slightly lower (average 33.8%) ( Figure 6 f).

[0206] Further comparison with ABE8e S109F 、ABE8e Y149V and ABE8e S109F / Y149V variant, as well as other previously reported ABE8e mutants, including ABE8e V106W 、ABE8e V82 、ABE8e K20A / R21 、ABE8e F148A 、ABE8e N108 and ABE8e N108Q / L145T The editing properties of (ABE9) in 102-sgRNA cells showed that ABE8e S109F and ABE8e Y149V Editing efficiency with ABE8e V106W 、ABE8e V82G and ABE8e K20A / R21A Equivalent, but with a narrower editing window ( Figure 6 g) In addition, although ABE8e S109F / Y149V The editing window is slightly wider than ABE8e N108Q or ABE9, but its editing efficiency at the A5 position was higher than these variants.

[0207] The above results show that the inventors' screening of 158 variants produced several ABE variants with high fidelity and high efficiency, among which ABE8e S109F 、ABE8e Y149V and ABE8e S109F / Y149V are optimal candidates for further off-target analysis.

[0208] Example 4, ABE8e Y149V and ABE8e S109F / Y149V Eliminated off-target effects genome-wide and transcriptome-wide

[0209] Considering ABE8e Y149V and ABE8e S109F Compared with ABE8e, it has a narrower editing window, lower bystander editing effect and comparable high target editing efficiency, and ABE8e S109F / Y149V With a stringent window and slightly lower efficiency, the inventors next examined whether the three variants exhibited reduced off-target effects in DNA and RNA by performing GOTI analysis in Ai9 mouse embryos and RNA sequencing (RNA-seq) in human HEK293T cells.

[0210] Since PCSK9 sgRNA has been used to lower cholesterol in mice and primates, the off-target effects of the three variants targeting this target site were evaluated using GOTI analysis. + The average targeting efficiency at the PCSK9 site in cells was high (87.4%-94.4%), and its off-target effect was significantly less than that of ABE8e ( Figure 7 a) In both single mutants, ABE8e S109F An average of 117 SNVs were generated per embryo, which was 70% lower than that of ABE8e, but still significantly higher than that of the Cre control group. Y149V Each embryo induced an average of only 24 SNVs, which was similar to the Cre control group. S109F / Y149V The double mutants introduced an average of 26 SNVs per embryo, which was also not significantly different from the Cre control group ( Figure 7 a). Further off-target SNV pattern and motif analysis also showed A-to-G mutation bias (accounting for 73% of total SNVs), with ABE8e S109F A slight preference for TA motifs was detected in editing, but in ABE8e Y149V or ABE8e S109F / Y149V There is no ( Figure 7 b and 7c).

[0211] These results confirm that ABE8e S109FStill, it causes relatively low genome-wide off-target effects, while ABE8e Y149V and ABE8e S109F / Y149V Compared with ABE8e, the DNA off-target effect was significantly reduced, reaching a level that was no different from the unedited control group.

[0212] To evaluate the off-target activities of these candidate ABEs at the RNA level, the inventors isolated ABE8e, ABE8e S109F 、ABE8e Y149V and ABE8e S109F / Y149V They were transfected into HEK293T cells with the same sgRNA, and the target site was HEK293-site 2. In addition, cells transfected with GFP and mCherry backbone plasmids were used as vector controls. After verifying the targeting efficiency of the four ABE variants by Sanger sequencing, RNA-seq (average depth 125×) analysis showed that ABE8e introduced an average of 11,396 RNA SNVs in the transcriptome, while 1,831 RNA SNVs were detected in the control group ( Figure 7 d). In addition, 93% of SNVs in the ABE8e group were A-to-G or U-to-C mutations, and were preferentially introduced at the TA motif ( Figure 7 e and 7f), indicating that ABE8e can mediate widespread adenine deaminase activity in the transcriptome of human cells. S109F (average 2543 SNVs), ABE8e Y149V (average 1646 SNVs) or ABE8e S109F / Y149V The number of RNA SNVs detected in the samples (average 1806 SNVs) was not significantly different from that in the control group. Similarly, no mutational bias or preferential recognition of motifs was observed compared with the control group ( Figure 7 e and 7f).

[0213] Taken together, these results suggest that ABE8e S109F The variants can reduce DNA off-target effects and eliminate RNA off-target effects, while ABE8e Y149V and ABE8e S109F / Y149V The variants can eliminate off-target effects in the genome and transcriptome. Y149V Compared to ABE8e S109F / Y149V With higher targeting efficiency, it is the relatively optimal candidate ABE8e variant.

[0214] Example 5, TadA8e Y149V Compatibility with different CRISPR systems

[0215] Considering ABE8e Y149VWith high editing efficiency and negligible off-target effects, the inventors have transformed TadA8e Y149V Combined with various CRISPR systems to expand its A sites that are not applicable to traditional NGG PAM. Since SpRY nuclease can target almost all PAMs, nSpCas9 was replaced with nSpRY to construct ABE8e-nSpRY and ABE8e Y149V -nSpRY base editors, and the activities of these two ABEs at 10 genomic sites with NNN PAMs (Table 2) were examined in HEK293T cells.

[0216] Targeted deep sequencing analysis showed that ABE8e-nSpRY mediated efficient A-to-G editing between A3-A10, with an average efficiency of more than 20% ( Figure 8 a). In contrast, ABE8e Y149V -nSpRY generally showed high editing activity at positions A5 and A6, with A-to-G base substitution efficiencies ranging from 9.52%-37.63% at NAN PAM sites, 33.65%-83.29% at NGN PAM sites, 38.37% at one NCN PAM site, and 26.38% at one NTN PAM site. The results obtained at these 10 target sites showed that ABE8e Y149V -nSpRY had a relatively higher average efficiency at position A6 than ABE8e-nSpRY (49.6% vs 42.5%), and its editing window was narrower (A4-A6; Figure 8 b) In addition, 7 of the 10 target sites contained at least one cytosine at positions 4–6, which were used to assess bystander C editing ( Figure 8 c) ABE8e at NNN positions 3 and 7 Y149V -nSpRY had significantly lower cytosine conversion activity than ABE8e-nSpRY (2.0% vs 9.2% and 1.2% vs 6.0%, respectively). Y149V -nSpRY C-editing activity was similar to that observed in the untreated vector control group.

[0217] In addition, the inventors constructed ABE8e-nSaKKH and ABE8e Y149V -nSaKKH editors were used to compare their activities at sites with NNNRRT PAMs (Tables 1-2). Testing at six genomic sites showed that ABE8e-nSaKKH had high efficiency (>50%) between positions A6 to A14. Y149V-nSaKKH showed high A-to-G substitution activity at positions A10 to A13, with an efficiency of 65.8%, slightly lower than that of ABE8e-nSaKKH ( Figure 8 d). In addition, ABE8e Y149V -nSaKKH's editing window (usually from A6 to A14) is narrower than ABE8e-nSaKKH's (from A2 to A17). Figure 8 e).

[0218] The inventors also transformed TadA8e Y149V Fused to the Cas9 ancestor IscB (496 amino acids), forming ABE8e-enIscB and ABE8e Y149V -enIscB microbase editors to compare their activities at sites with NWR RNA TAMs (Tables 1-2). Y149V -enIscB showed the highest A-to-G conversion efficiency at position A4 (27.2% on average), which was comparable to that of ABE8e-enIscB (28.6% on average). Y149V -enIscB has a relatively lower A editing activity than ABE8e-enIscB ( Figure 8 f and 8g).

[0219] Overall, these results suggest that TadA8e Y149V Compatible with PAM-relaxed Cas9 variants, ultra-compact CRISPR, and ancestral Cas9 systems, thereby expanding the target range for efficient adenine base editing.

[0220] Example 6, ABE8e Y149V Precise editing of Hpd in vivo to treat fatal inherited tyrosinemia type I To explore the role of ABE8e Y149V The inventors selected Fah for its potential in gene therapy for treating genetic diseases - / -mice, a widely used model of a fatal genetic disease, hereditary tyrosinemia type I (HTI). HTI is caused by loss-of-function mutations in fumarylacetoacetate hydrolase (FAH), an enzyme in the tyrosine metabolic pathway. HTI is characterized by abnormal accumulation of tyrosine and its metabolites, leading to hepatotoxicity, ultimately damaging the liver and causing weight loss. By inhibiting the upstream enzyme hydroxyphenylpyruvate dioxygenase (HPD) using nitisinone, the accumulation of this tyrosine metabolite can be blocked and fatal liver failure can be prevented. Previous studies have reported that HTI mice can be treated by Cas9 nuclease-mediated disruption of the Hpd gene and CBE-mediated introduction of a premature stop codon in the Hpd gene. Here, we aimed to adopt an alternative gene silencing strategy, as previously described, by ABE8e Y149V mediated in vivo start codon mutation to disrupt the Hpd gene ( Fig. 9 a).

[0221] Due to the limited loading capacity of AAV (less than 5 kb), a dual AAV system was used and ABE8e Y149V It is divided into two parts, which are fused to the N-terminus and C-terminus of Cfa intein respectively ( Fig. 9 b) Both constructs were packaged into AAV8 serotype and delivered via tail vein injection into 6- to 8-week-old Fah - / - In mice, saline injection was used as Fah - / - After injection, mice continued to take Nitisinone for 7 days and then stopped taking it ( Fig. 9 b) At day 32 after injection, mice in the control group had lost less than 20% of their body weight and were euthanized. In contrast, mice in the treated group behaved like wild-type mice; they gained weight and did not require nitisinone ( Fig. 9 c).

[0222] The editing efficiency and liver function of mice were tested on day 30. Y149V In treated mice, deep sequencing of the Hpd gene in liver tissue showed that the average base editing efficiency of the target A to G precise mutation at position A6 was 69.0%, and there was almost no bystander A editing, C editing, and indel editing ( Fig. 9 d). In addition, immunohistochemical (IHC) staining of liver sections using anti-HPD antibody detected extensive HPD-negative hepatocyte patches ( Fig. 9 e).

[0223] Serum biomarkers including aspartate aminotransferase (AST), alanine aminotransferase (ALT), and total bilirubin levels showed that ABE8e Y149V Processing Fah - / -The liver function of mice (not taking nitisinone) was significantly improved, compared with that of Fah mice taking nitisinone. - / - The results were similar to those of mice, but significantly better than those of Fah mice injected with saline. - / - Mice (not treated with nitisinone) Fig. 9 fh).

[0224] These results indicate that ABE8e Y149V It has the characteristics of efficient and precise gene editing in vivo and is a meaningful base editor for human genetic diseases.

[0225] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for reducing the off-target effect of an adenine base editor and improving the editing accuracy, comprising: TadA8e is used as adenine deaminase, and the amino acid residues of TadA8e are modified; the modification includes mutating the amino acid residues Y149, S109, N119, N122, D147, R111, I166 or N167 of TadA8e.

2. The method according to claim 1, characterized in that The modification of the amino acid residue includes: mutating the amino acid residue into other amino acid residues other than the residue; Preferably, the modification of the amino acid residues includes: Y149V, Y149L, Y149M or Y149I for the Y149 site; S109F for the S109 site; N119C, N119Q or N119D for the N119 site; N122F, N122V, N122L or N122A for the N122 site; and D147K for the D147 site. Preferably, the modification of the amino acid residues includes: Y149V and / or S109F.

3. A TadA8e variant comprising a mutation of an amino acid residue selected from the group consisting of: Y149, S109, N119, N122, D147, R111, I166 or N167; preferably, the mutation comprises: For Y149V, Y149L, Y149M or Y149I at the Y149 site; for S109F at the S109 site; for N119C, N119Q or N119D at the N119 site; for N122F, N122V, N122L or N122A at the N122 site; and / or; for D147K at the D147 site; more preferably, the mutation includes: Y149V and / or S109F.

4. An adenine base editor, comprising the TadA8e variant of claim 3, and an element suitable for adenine editing operably connected thereto; preferably, the adenine base editor comprises: a TadA8e variant, a Cas nuclease operably connected thereto; More preferably, the adenine base editor comprises operably linked: NLS, TadA8e variant, Cas nuclease, NLS; More preferably, the Cas nuclease includes: Cas9 nuclease or its homologs, and the homologs include: SpRY nuclease, SaKKH nuclease, IscB nuclease.

5. An isolated polynucleotide encoding: The TadA8e variant of claim 2; or The adenine base editor of claim 4.

6. A construct or an expression vector containing the construct, which contains the polynucleotide according to claim 5; Preferably, the construct comprises the expression cassette of the adenine base editor according to claim 4, and an operably connected sgRNA expression cassette targeting the target gene; Preferably, the target gene includes: Hpd, the construct or the expression vector containing the construct is used to prepare a pharmaceutical composition for alleviating or treating hereditary type I tyrosinemia; More preferably, the nucleotide sequence of the sgRNA is as shown in SEQ ID NO:

7.

7. The construct according to claim 6 or an expression vector containing the construct, characterized in that: The construct comprises: Construct 1: an expression cassette for the sgRNA targeting the target gene, a gene encoding a TadA8e variant, and a gene encoding the N-terminal portion of the Cas nuclease; and Construct 2: Expression cassette for the C-terminal part of the Cas nuclease.

8. A recombinant cell comprising the construct or expression vector according to claim 6 or 7, or the polynucleotide according to claim 5 integrated into its genome.

9. Use of the adenine base editor of claim 4 or the construct of claim 6 or 7 or an expression vector containing the construct for gene editing or for preparing a reagent for gene editing; wherein the gene editing has reduced off-target effects and improved editing accuracy; preferably, the gene editing is adenine base editor-mediated gene editing; preferably, the gene editing is conversion of A·T to G·C.

10. A method for gene editing, comprising mediating gene editing using the adenine base editor of claim 4 or the construct of claim 6 or 7 or an expression vector containing the construct; Preferably, the nucleic acid sequence encoding the adenine base editor is transfected or injected into a receptor to perform gene editing; preferably, the receptor comprises: Somatic or germ cells; Preferably, the reproductive cells include embryonic cells or fertilized eggs.

11. The use according to claim 9 or the method according to claim 10, characterized in that The gene editing includes gene editing targeting the target gene Hpd; preferably, the reagent for gene editing is used to prepare a pharmaceutical composition for alleviating or treating hereditary type I tyrosinemia; preferably, in the gene editing targeting the target gene Hpd, the nucleotide sequence of sgRNA is as shown in SEQ ID NO:

7.

12. A reagent or kit for gene editing, comprising: The TadA8e variant of claim 3, or a polynucleotide encoding the same; or The adenine base editor of claim 4, or a polynucleotide encoding the same; or The construct according to claim 6 or 7 or an expression vector containing the construct.

Images