WO2020241869A1 - GENOME EDITING SYSTEM USING Cas PROTEIN HAVING TWO TYPES OF NUCLEIC ACID BASE-CONVERTING ENZYMES FUSED THERETO - Google Patents

Abstract

According to the present invention, a fusion protein has different nucleic acid base-converting enzymes which are respectively bonded to an N-terminal and a C-terminal of a Cas protein, wherein the Cas protein partially or fully lacks nuclease activity.

Description

Genome editing system using Cas protein, which is a fusion of two types of nucleobase converting enzymes

The present invention relates to a genome editing system using a Cas protein in which two types of nucleic acid-based converting enzymes are fused. More specifically, the present invention is a fusion protein in which different nucleic acid-based converting enzymes are bound to the N-terminal and C-terminal of the Cas protein. With respect to fusion proteins in which the Cas protein has lost some or all of its nuclease activity. The present invention also relates to a polynucleotide encoding the fusion protein, an expression vector containing the polynucleotide, a genome editing system containing the polynucleotide, and a method for producing a DNA-edited cell using the system.

The genome, which is the blueprint for life, is composed of a DNA base sequence consisting of the four letters A, T, G, and C. In recent years, genome editing technology that can freely edit a specific target sequence in intracellular genomic DNA has been rapidly developed, and is bringing about a great change in the entire biology field including agriculture and medical fields.

In particular, in CRISPR-Cas9 genome editing, Cas9Cas9 protein having DNA cleaving activity is recruited by a guide RNA (gRNA) to a target DNA region located upstream of the 3'protospacer adjacent motif (PAM), and double-stranded DNA. It is a genome editing tool that induces cleavage (DSB) (Non-Patent Documents 1 and 2). This method has been widely applied as a genome editing tool, such as knocking out a specific gene through error-prone DNA repair and enabling insertion of another foreign DNA into a chromosome through DSB-induced homologous recombination. There was. However, in the CRISPR-Cas9 genome editing method, cytotoxicity due to double-stranded DNA cleavage and editing accuracy have been problems.

Under these circumstances, for the past few years, a base editing tool has been developed in which a deaminase (deaminase) is fused with a mutant Cas9 that does not have DNA cleaving activity. More specifically, base editing induced by nuclease deficiency or tethering deoxynucleoside deaminase to a complex consisting of nickase Cas9 (dCas9 or nCas9) and gRNA is an efficient and direct base in the genomic sequence. It has been clarified that it induces substitution (Non-Patent Document 3).

Among the currently available base edits, cytosine base edits (CBEs, Non-Patent Documents 4, 5 and 7) and adenine base edits (ABEs, Non-Patent Documents 6 and 7) are performed in a narrow region of the gRNA target site. Allows for very efficient and accurate base substitution.

CBEs are usually composed of cytidine deaminase and uracil glycosylase inhibitor (UGI). The cytidine deaminase is rAPOBEC1 used in BEs (Non-Patent Document 5), PmCDA1 used in Target-AID (Non-Patent Document 4), etc., and converts cytidine in a non-gRNA-binding DNA strand into uridine. UGI also inhibits base excision repair. Then, by using these, it becomes possible to replace cytosine with uracil and further with thymine through DNA repair.

In ABEs, for example, a heterodimer complex consisting of wild-type and modified TadA adenosine deaminese is used. The enzyme converts adenine to inosine and replicates it as guanine (Non-Patent Document 6).

In this way, in base editing, C of the target DNA can be converted to T or A to G with high efficiency (C → T or A → G) through deamination reaction without cutting the DNA. Expectations are rising as an excellent genome editing tool in areas such as crop breeding and gene therapy. However, the base substitution patterns that can be used with conventional base editing tools have been limited to two types, C → T or A → G.

The present invention has been made in view of the above-mentioned problems of the prior art, and an object of the present invention is to provide a base editing tool capable of simultaneously performing a plurality of types of base substitutions.

The present inventors aim to provide a base editing tool that enables simultaneous base substitution of both C → T and A → G in order to achieve the above object, and cytidine deaminase and adenosine. Adenosine was fused to a single nCas9 (D10A) to first prepare three base editors, Target-ACE, Target-ACEmax, and ACBEmax.

In Target-ACE, nCas9 fused to PmCDA1 derived from Target-AID and TadA heterodimer derived from ABE-7.10 are fused as C-terminal and N-terminal, respectively, for consistent single-function base editing. It is configured with other functional domains that exist.

Target-ACEmax was constructed by applying GenScript codon optimization and addition of a binode-type nuclear localization signal (binode-type NLS) to the N-terminal to the Target-ACE.

ACBEmax was constructed by replacing the codon-optimized PmCDA1 domain in Target-ACEmax with the codon-optimized cytidine deaminase domain rAPOBEC1 derived from BE4max.

Next, using a conventional single-function base editor, a combination thereof, the above-mentioned Target-ACEmax, etc., genome editing of 47 target regions on the genome in cultured human cells was performed, and these base substitution patterns were subjected to amplicon. Observed by sequencing. As a result, it was clarified that all of Target-ACE, Target-ACEmax and ACBEmax have high co-editing efficiency of different bases. In particular, it was revealed that Target-ACEmax has the highest co-editing efficiency of heterologous bases.

In addition, regarding the editing of unintended genomic regions (DNA off-target effect), which is generally regarded as a problem in genome editing technology, and the editing of unintended intracellular RNA molecules (RNA off-target effect), which is a concern in base editing technology, exome. Analysis and transcriptome analysis were performed. As a result, it was also revealed that Target-ACEmax does not unnecessarily enhance these off-target effects as compared with conventional base editing tools.

We also developed a machine learning model that predicts the base editing pattern of an arbitrary target sequence based on the huge amount of base editing pattern evaluation data obtained by amplicon sequencing. Using this machine learning model, it is possible to accurately induce various amino acid substitutions in the gene group in which Target-ACEmax encodes a protein by suppressing unnecessary base editing other than the target base group, and various observed in humans. It has also been shown that a plurality of disease gene mutations can be repaired at the same time with high efficiency, and the present invention has been completed.

That is, the present invention relates to a fusion protein in which different nucleobase converting enzymes are bound to the N-terminal and C-terminal of the Cas protein, and the Cas protein loses part or all of the nuclease activity. The present invention also relates to a polynucleotide encoding the fusion protein, an expression vector containing the polynucleotide, a genome editing system containing the polynucleotide, and a method for producing a DNA-edited cell using the system. Specifically, it provides the following.
<1> A fusion protein in which different nucleobase converting enzymes are bound to the N-terminal and C-terminal of the Cas protein, and the Cas protein loses part or all of its nuclease activity.
<2> The fusion protein according to <1>, wherein the different nucleobase converting enzymes are adenosine deaminase and cytidine deaminase.
<3> The fusion protein according to <2>, wherein the adenosine deaminase is bound to the N-terminal of the Cas protein and the cytidine deaminase is bound to the C-terminal of the Cas protein.
<4> The fusion protein according to any one of <1> to <3>, to which at least one of a nuclear localization signal and a uracil glycosylase inhibitor is bound.
<5> A polynucleotide encoding the fusion protein according to any one of <1> to <4>.
<6> The polynucleotide according to <5>, wherein the codon is optimized according to the host cell to be introduced.
<7> An expression vector containing the polynucleotide according to <5> or <6>.
<8> A genome editing system including the following (A) and (B).
(A) The fusion protein according to any one of <1> to <4>, the polynucleotide according to <5> or <6>, or the expression vector according to <6>.
(B) A method for producing a guide RNA, a polynucleotide encoding the guide RNA, or an expression vector <9> DNA-edited cell containing the polynucleotide, wherein the cell is subjected to the genome according to <8>. Methods involving the introduction of an editing system.

According to the present invention, it is possible to simultaneously perform a plurality of types of base substitutions with one base editing tool.

It is a schematic diagram which shows the development system of the dual-function base editor (Target-ACEmax, Target-ACE, ACBEmax) of this invention. The development line from the single-function base editor to the dual-function base editor is indicated by an arrow, and the combination of single-function base editors corresponding to the dual-function base editor is indicated by a broken line. It is a figure which shows the outline of the structure of Target-ACEmax. The schematic diagram of the C → T base editing reporter system is shown. Following C → T base editing on the antisense strand, DNA replication restores the translation of EGFP by converting the codon GTG (coding valine) to the start codon ATG (coding methionine). The schematic diagram of the A → G base edit reporter system is shown. Following A → G base editing in the antisense strand, DNA replication converts the stop codon TAA to codon CAA (encoding glutamine), which proceeds with the translation of EGFP downstream thereof. Photographs of C → T and A → G base editing reporter cells transiently introduced with various base editors and corresponding on-target (OT) or non-target (NT) gRNAs observed with a fluorescence microscope. .. The scale bar in the figure represents 40 μm. Consistent results were independently reproduced in 4 cell cultures. It is a graph which shows the restoration frequency of a start codon in a C → T base edit reporter cell. In the figure, the results of three independent transfection experiments are indicated by dots, and the average value thereof is indicated by a bar. It is a graph which shows the disruption frequency of a stop codon in A → G base edit reporter cells. In the figure, the results of three independent transfection experiments are indicated by dots, and the average value thereof is indicated by a bar. FIG. 5 is a graph showing the frequency with which C → T base editing was detected by amplicon sequencing at the gRNA target site (-30 bp to +10 bp with respect to PAM) of the C → T base editing reporter cell. In the figure, the results of three independent transfection experiments are indicated by dots, and the average value thereof is indicated by a bar. FIG. 5 is a graph showing the frequency with which A → G base editing was detected by amplicon sequencing at the gRNA target site (-30 bp to +10 bp with respect to PAM) of the A → G base editing reporter cell. In the figure, the results of three independent transfection experiments are indicated by dots, and the average value thereof is indicated by a bar. It is a graph which shows the mean C-> T and A-> G base edit spectrum at a genome target site. It is a graph which shows the editing frequency of C → T or A → G at 47 endogenous target sites of the human genome. The black horizontal bars represent the average edit spectrum of C → T or A → G. A bilateral Mann-Whitney U test was performed to compare any pair in the two datasets obtained from three cell cultures. Arrows indicate datasets that show a higher average editing frequency than others (* P <0.05; ** P <0.01; *** P <0.001). It is a figure which shows the editing pattern by the co-editing of C → T and A → G. A total of 722 heterogeneous co-editing patterns obtained by base editing conditions with a read frequency threshold of 0.1% (upper panel) are hierarchical based on the frequencies generated by various base editing conditions (lower panel). Clustered in. The total number of edit result patterns observed in the three independent repeat experiments is shown on the right side of each base edit condition. It is a figure which shows the average frequency of C-> T and A-> G simultaneous editing which occurs in the -20 to -1bp upstream region of PAM. It is a graph which shows the co-editing frequency by a mixture of a dual-function base editor or a single-function base editor in the combination of cytosine and adenine located at a specific site with respect to PAM. The combination of positions in which the average co-editing frequency is ranked in the top 5 under any base editing condition is shown. Each bar shows the mean co-editing frequency measured for cytosine and adenine-bearing target sites at each combination position in three cell cultures. A two-sided Mann-Whitney U test was performed to compare Target-ACEmax with the corresponding enzyme mixture and the other two bibase editors for mean co-editing frequency in a combination of positions with sufficient sample size. (* P <0.05; ** P <0.01; *** P <0.001). It is a graph which shows the base edit frequency in the EMX1 gene site 1, the FANCF gene site 1 and site 2, and the corresponding off-target site. The ampli-consequencing experiment was performed 3 times. It is a graph which shows the result of estimating the DNA off-target risk. The jitter plot shows the relative off-target / on-target edit frequency measured for the three gRNAs in three cell culture reproduction experiments, and the black horizontal bar represents the median DNA off-target risk score. The area around 10 ^-4 to 10 ^-2 (the area in pink under the color display) indicates the score range of the single-function base editor. A two-sided Welch's t-test was performed and a dual-function editing editor was compared with the corresponding single-function enzyme mixture. Arrows indicate datasets with higher mean scores than others (** P <0.01). It is a graph which shows the number and frequency of genome-wide C → U RNA mutants detected in the cell subjected to various base editing conditions. Each bar shows the number of mutants identified by RNA-seq, and the jitter plot shows their mutant aryl frequency. The experiment was performed twice for each base editing condition. It is a graph which shows the number and frequency of genome-wide A → I RNA mutants detected in the cell subjected to various base editing conditions. Each bar shows the number of mutants identified by RNA-seq, and the jitter plot shows their mutant aryl frequency. The experiment was performed twice for each base editing condition. It is a figure which shows the outline of the conditional probability model in order to predict the frequency of various base editing patterns in an arbitrary target DNA sequence by various base editing editors. The dot diagram at the bottom shows the correlation between the measured value and the predicted value. It is a graph which shows the bystander mutation risk score with codon conversion in various base editing methods. The horizontal bar represents the median risk score for the various codon conversion types represented by the dots. 10 ⁰ ~ 4 × 10 ⁰ near region (color display under pink areas) indicates the range of the average risk score of single-function base editor except BE4max (C). A two-sided Mann-Whitney U test was performed to compare any pair in the two datasets. Arrows indicate datasets with higher mean scores than others (*** P <0.001). It is a graph which shows the frequency of the heterogeneous disease mutation pair which can be modified. In the case of pathological C / G → TA and A / T → G / C single nucleotide polymorphisms (SNVs) reported in the ClinVar database, the gRNA target sites that may correct each mutation are from the target mutation. It was first screened within the ± 25 bp region. A base-edited predictive model trained by all amplicon sequencing datasets to determine the probability of correcting mutations at these different gRNA target sites without inducing unwanted bystander mutations in the ± 15 bp region from the target mutations. Predicted using. For each mutation, its modification rate was defined as the maximum probability of a target conversion pattern among those induced by various RNAs. Finally, the ability of heterologous base editing conditions to correct heterologous mutations was evaluated by counting the combination of two heterologous mutations in the same disease gene, both of which could be corrected with a 5% correction rate threshold. I predicted.

<Fusion protein>
The present invention provides a fusion protein in which different nucleobase converting enzymes are bound to the N-terminal and C-terminal of the Cas protein, wherein the Cas protein loses part or all of its nuclease activity.

The Cas protein in the fusion protein of the present invention has lost part or all of its nuclease activity. A Cas protein that has lost part of its nuclease activity is called "nCas", and a Cas protein that has lost all of its nuclease activity is called "dCas". The Cas protein typically comprises a domain involved in cleavage of the target strand (RuvC domain) and a domain involved in cleavage of the non-target strand (HNH domain), but the Cas protein used in the present invention is preferably preferred. The introduction of a mutation into at least one domain results in the loss of nuclease activity in that domain. As such a mutation, in the case of SpCas9 protein (Cas9 protein derived from S. pyogenes), for example, a mutation of the 10th amino acid (aspartic acid) from the N-terminal to alanine (D10A: mutation in the RuvC domain). , Mutation of amino acid 840th from N-terminal (histidine) to alanine (H840A: mutation in HNH domain), mutation of amino acid 863th from N-terminal (aspartic acid) to alanine (N863A: mutation in HNH domain) , The mutation of the 762th amino acid (glutamic acid) from the N-terminal to alanine (E762A: mutation in the RuvCII domain), the mutation of the 986th amino acid (aspartic acid) from the N-terminal to alanine (D986A: mutation in the RuvCIII domain) ). In addition, Cas9 proteins of various origins are known (eg, WO2014 / 131833), and their nCas or dCas can be utilized. The amino acid sequence and base sequence of Cas9 protein are registered in a public database, for example, GenBank (http://www.ncbi.nlm.nih.gov) (for example, accession number: Q99ZW2.1, etc.). ), These can be used in the present invention.

Further mutations, such as mutations to alter PAM recognition, may be introduced into the Cas protein (Benjamin, P. et al., Nature 523, 481-485 (2015), Hirano, S. et al., Molecular Cell. 61, 886-894 (2016)).

In the present invention, Cas proteins other than Cas9, for example, Cpf1 (Cas12), Cas12b, CasX (Cas12e), Cas14 and the like can also be used.

In the present invention, the "nucleobase converting enzyme" refers to a target nucleotide without breaking the DNA strand by catalyzing a reaction of converting a substituent on the purine or pyrimidine ring of a DNA base into another group or atom. Means an enzyme capable of converting a nucleic acid into another nucleotide, and examples thereof include deaminase. A "deaminase" is an enzyme that catalyzes a deamination reaction that converts an amino group of a base into a carbonyl group and belongs to the nucleic acid / nucleotide deaminase superfamily. Examples of such deaminase include cytosine deaminase capable of converting cytosine or 5-methylcytosine to uracil or thymine, adenosine deaminase capable of converting adenine to hypoxanthine, and guanosine deaminase capable of converting guanine to xanthine. The origin of the nucleobase converting enzyme is not particularly limited, and examples thereof include mammals (eg, humans, pigs, cows, horses, monkeys), fish (eg, lampreys), and bacteria (eg, Escherichia coli).

The "different nucleobase converting enzyme" fused to the N-terminal and C-terminal of the Cas protein is not particularly limited, but a combination of adenosine deaminase and cytidine deaminase is preferable. Further, in the fusion protein, it is preferable that adenosine deaminase is bound to the N-terminal of the Cas protein and cytidine deaminase is bound to the C-terminal of the Cas protein.

Examples of adenosine deaminase include Escherichia coli-derived TadA and its variants. A typical amino acid sequence of wild-type TadA includes an amino acid sequence consisting of positions 218 to 383 shown in SEQ ID NO: 2, and an amino acid sequence of a variant thereof (TadA *) is shown in SEQ ID NO: 2. Examples thereof include the amino acid sequences consisting of the 20th to 185th positions described above. In addition, the fusion protein of the present invention has both TadA and a variant thereof from the viewpoint of facilitating reduction of off-target RNA editing while maintaining the activity of on-target DNA base substitution. desirable.

Examples of the cytosine deaminase include PmCDA1 (Petromyzon mammal cytosine deaminase 1), APOBEC (APOBEC1, APOBEC2, APOBEC3 (APOBEC3A, 3B, 3C, 3D (3E), etc.), 3G, 3D (3E), 3F, 3F, etc. Examples thereof include Anc689, which is an ancestral amino acid sequence of APOBEC, AID (Activation-induced cytosine deaminase (AICDA)) derived from mammals, a family of AIDs, and variants thereof. Among these, PmCDA1 is preferable from the viewpoint that it easily exhibits high catalytic activity when bound to the C-terminal of Cas protein. A typical amino acid sequence of PmCDA1 includes the amino acid sequence consisting of positions 1876 to 2083 shown in SEQ ID NO: 2. Examples of the amino acid sequence of APOBEC1 include the amino acid sequence at positions 1876 to 2103 shown in SEQ ID NO: 6.

It is preferable that at least one of the nuclear localization signal and the uracil glycosylase inhibitor is further bound to the "fusion protein" of the present invention.

The "nuclear localization signal" according to the present invention is not particularly limited as long as it is an amino acid sequence that serves as a marker for transporting the fusion protein of the present invention to the cell nucleus, but is a sequence having one cluster consisting of basic amino acids ( It may be a mononode type) or a sequence in which clusters consisting of two basic amino acids are linked via a spacer sequence (binode type). The nuclear localization signal may be bound to at least one of the N-terminal and the C-terminal in the fusion protein of the present invention, or may be inserted between each domain.

The "uracil glycosylase inhibitor" according to the present invention may be any as long as it can suppress the degradation of uracil converted from cytosine by cytidine deaminase by uracil DNA glycosylase, for example, 2095 described in SEQ ID NO: 2. An amino acid sequence consisting of ~ 2177 positions can be mentioned.

The uracil glycosylase inhibitor may be bound to at least one of the N-terminal and the C-terminal in the fusion protein of the present invention as long as it exhibits the inhibitory activity, or may be inserted between the domains. However, it is preferable that it is bound to the C-terminal. Further, the number of uracil glycosylase inhibitors in the fusion protein of the present invention is not particularly limited as long as the inhibitory activity can be exhibited, and may be one, or a plurality (for example, 2, 3, 4, or 5). It may be (pieces).

In the fusion protein of the present invention, the above-mentioned Cas protein, nucleobase converting enzyme, nuclear translocation signal and uracil glycosylase inhibitor are, for example, adenosine deaminase-Cas protein-citidine deaminase-uracil glycosylase inhibition in order from N-terminal as follows. Placed with the agent.

In addition, as a preferred embodiment of the fusion protein of the present invention, nuclear localization signal-adenosin deaminase-Cas protein-cytidine deaminase-nuclear transfer signal-uracil glycosylase inhibitor can be mentioned in order from N-terminal, and more preferably N-terminal. In order from, binode nuclear localization signal-modified TadA (TadA *) -TadA-SpCas9 (D10A) -PmCDA1-mononode nuclear localization signal-uracil glycosylase inhibitor is mentioned, more preferably SEQ ID NO: 2. It is a protein (Taget-ACEmax) consisting of the amino acid sequence described in 1.

Another preferred embodiment of the fusion protein of the present invention includes adenosine deaminase-Cas protein-citidine deaminase-nuclear localization signal-uracil glycosylase inhibitor in order from N-terminal, and more preferably modified in order from N-terminal. Type TadA (TadA *)-TadA-SpCas9 (D10A) -PmCDA1-mononode nuclear localization signal-uracil glycosylase inhibitor, more preferably a protein consisting of the amino acid sequence set forth in SEQ ID NO: 4 (Tadet-). ACE).

A preferred embodiment of the fusion protein of the present invention includes, in order from the N-terminal, a nuclear localization signal-adenosine deaminase-Cas protein-citidine deaminase-nuclear transfer signal-uracil glycosylase inhibitor-nuclear transfer signal, and more preferably From the N-terminal, binode nuclear localization signal-modified TadA (TadA *) -TadA-SpCas9 (D10A) -APOBEC1-urasyl glycosylase inhibitor-urasyl glycosylase inhibitor-binode nuclear localization signal, and more. Preferably, it is a protein (ACBEmax) consisting of the amino acid sequence shown in SEQ ID NO: 6.

Further, in the fusion protein of the present invention, the Cas protein, nucleobase converting enzyme, nuclear localization signal and uracil glycosylase inhibitor described above may be directly bound or indirectly bound via a linker sequence. May be good.

In the present invention, the "linker sequence" is not particularly limited as long as the function of each of the proteins is not suppressed, and a linker composed of glycine and serine (GGS peptide, 2 × GGS peptide, 3 × GGS peptide, GS peptide, 2 × GS peptide, 3 × GS peptide, etc.). In addition, only one of these linker sequences may be arranged between each protein, or a plurality of (for example, 2, 3, 4 or 5 sequences) may be arranged. When arranging a plurality of linker sequences, it may be composed of only one kind of linker sequence, or may be composed of a plurality of kinds of linker sequences.

In the present invention, the number of amino acids in the "linker sequence" arranged between each protein is not particularly limited as long as the function of each protein is not suppressed, but is usually 1 to 300 amino acids, and the lower limit is set. It is preferably 2 amino acids or more (for example, 3 amino acids or more, 4 amino acids or more, 5 amino acids or more, 6 amino acids or more, 7 amino acids or more, 8 amino acids or more, 9 amino acids or more), and more preferably 10 amino acids or more (for example, 20 amino acids or more). Amino acids or more, 30 amino acids or more, 40 amino acids or more, 50 amino acids or more, 60 amino acids or more, 70 amino acids or more, 80 amino acids or more, 90 amino acids or more). The upper limit is preferably 200 amino acids or less (for example, 190 amino acids or less, 180 amino acids or less, 170 amino acids or 160 amino acids), and more preferably 150 amino acids or less (for example, 140 amino acids or less, 130 amino acids or less, 120 amino acids). Hereinafter, it is 110 amino acids or less, more preferably 100 amino acids or less (for example, 90 amino acids or less, 80 amino acids or less, 70 amino acids or less, 60 amino acids or less), and more preferably 50 amino acids or less (for example, 40 amino acids or less, 30 amino acids or less, 20 amino acids or less, 10 amino acids or less).

Further, the fusion protein of the present invention may have other functional proteins in addition to the above-mentioned Cas protein, nucleobase converting enzyme, nuclear translocation signal, uracil glycosylase inhibitor and linker sequence. The other functional protein is not particularly limited, and is appropriately selected according to the function to be imparted to the fusion protein of the present invention. For example, functional proteins used for the purpose of facilitating purification and detection of fusion proteins include, for example, 3 × FLAG tag peptide, FLAG tag peptide (both registered trademarks, Sigma-Aldrich), XTEN peptide, SH3 peptide and the like. Examples thereof include modified versions, Myc tags, His tags, HA tags, and fluorescent protein tags (GFP and the like).

Although the fusion protein of the present invention has been described above, specific amino acid sequences relating to various proteins constituting the fusion protein can be found in other than those shown above, as well as known literature and NCBI (http://https) by those skilled in the art. : //Www.ncbi.nlm.nih.gov/guide/) and the like can be searched and obtained as appropriate. Further, not limited to the typical amino acid sequence (for example, NCBI reference sequence) obtained in this way, the proteins and the like according to the present invention are the typical amino acids as long as each function of the protein is maintained. It may also include variants and homologues to the sequence. In addition, such variants and homologues usually have high homology to a typical amino acid sequence. The high homology is usually 60% or more, preferably 70% or more, more preferably 80% or more, still more preferably 90% or more (eg, 95% or more, 96% or more, 97% or more, 98% or more, 99% or more).

In addition, the fusion protein of the present invention can be obtained by a person skilled in the art based on the information of the nucleotide sequence encoding the amino acid sequence of Escherichia coli, animal cells, insect cells, plant cells, and cell-free protein synthesis system (for example, reticulated red erythrocyte extraction). It can also be biosynthesized by a genetic method using a liquid, wheat germ extract) or the like. Further, based on the information on the amino acid sequence, it can be chemically synthesized using a peptide synthesizer or the like.

<Polynucleotides and vectors>
The present invention provides polynucleotides encoding the fusion proteins described above. The polynucleotide is not particularly limited, but it is preferable to optimize the codons used over the entire length of the CDS according to the host cell to be introduced. When expressing a heterologous polynucleotide, the protein expression level can be expected to increase by converting the sequence into codons that are frequently used in the host organism. For example, the codon usage frequency data in the host used can be obtained from the genetic code usage frequency database (http://www.kazusa.or.jp/codon/index.html) published on the homepage of the Kazusa DNA Research Institute. ) Can be used, or the literature describing the frequency of codon usage in each host may be referred to. By referring to the obtained data and the DNA sequence to be introduced, the codons used in the DNA sequence that are not frequently used in the host can be converted into codons that are frequently used by encoding the same amino acid. Good. In addition, it can be optimized by using programs provided by various manufacturers. Examples of such manufacturers and programs provided include GenScript (OptimumGene), Integrated DNA Technologies (Codon Optimization), and Eurofins Genomics (GENEius), among which GenScript codon (OptimumGene) by GenScript codon (Op) Is preferable.

The present invention may also take the form of an expression vector such that the fusion protein encoded by the polynucleotide can be expressed in the host. When adopting the form of an expression vector, it comprises one or more regulatory elements that are operably linked to the polynucleotide (DNA) to be expressed. Here, "operably bound" means that the DNA is operably bound to the regulatory element. "Regulatory elements" include promoters, enhancers, internal ribosome entry sites (IRES), self-cleaving peptides (eg, self-cleaving 2A peptides), and other expression control elements (eg, transcription termination signals, eg, polyadenylation). Signal and poly U sequences). Regulatory elements, depending on the purpose, for example, those that direct the constitutive expression of DNA in various host cells but that direct the expression of DNA only in specific cells, tissues, or organs. It may be. Further, the expression of DNA may be directed only at a specific time, or the expression of artificially inducible DNA may be directed. Examples of promoters include the polII promoter (eg, retrovirus Rous sarcoma virus (RSV) LTR promoter, cytomegalovirus (CMV) promoter, SV40 promoter, dihydrofolate reductase promoter, β-actin promoter, phosphoglycerol kinase (PGK)). Promoters and EF1α promoters), polI promoters, or combinations thereof. In addition, the expression vector of the present invention may contain a selectable marker (drug resistance gene, nutritional requirement complementary gene, etc.) so that introduction into a host cell can be detected.

A person skilled in the art can select an appropriate expression vector according to the type of cell to be introduced and the like. In addition, such an expression vector chemically synthesizes a DNA strand, or, as shown in Examples described later, a partially overlapping oligo DNA short strand synthesized by using a PCR method or a Gibson Assembly method. By connecting, it is also possible to construct a DNA encoding the entire length.

<Genome editing system>
The present invention provides a genome editing system including the following (A) and (B).
(A) The fusion protein, a polynucleotide encoding the protein, or an expression vector containing the polynucleotide.
(B) A guide RNA, a polynucleotide encoding the guide RNA, or an expression vector containing the polynucleotide.

(A) is as described above. Regarding (B), the guide RNA is a combination of crRNA and tracrRNA corresponding to the CRISPR / Cas9 system. The crRNA and tracrRNA may be in the form of a single molecule (for example, a form in which an intervening sequence is sandwiched between them) or a two-molecule form. The crRNA contains at least a base sequence complementary to the base sequence of the target DNA region (targeted base sequence) and a base sequence capable of interacting with tracrRNA in this order from the 5'side. The crRNA forms a double-stranded RNA with the tracrRNA in a base sequence capable of interacting with the tracrRNA, and the formed double-stranded RNA interacts with the Cas9 protein. This guides the Cas9 protein to the target DNA region.

In the target DNA region, base editing at the target site is performed within the sequence on the target DNA of the guide RNA (guide RNA target sequence) located upstream (5'side) of the PAM sequence recognized by the Cas9 protein. Occurs. More specifically, the complementarity of base pairing between the guide RNA target sequence and the targeting base sequence which is a complementary sequence thereof, and the PAM sequence located on the 3'side of the complementary strand of the guide RNA target sequence. And, it happens at the position determined by both. The guide RNA of the Cas9 protein according to the present invention recognizes the complementary strand of the base sequence existing on the 5'side of the PAM sequence so that the PAM sequence can be recognized by the Cas9 protein, that is, a guide. The guide RNA target sequence recognized by RNA is designed to be the base sequence of the complementary strand of the base sequence on the 5'side of the PAM sequence so that the PAM sequence can be recognized by the Cas9 protein.

Various methods are known as methods for designing the PAM sequence and guide RNA target sequence of such Cas9 protein. For example, CRISPR Design Tool (http://crispr.mit.edu/) (Massachusetts Institute of Technology). ), E-CRISPR (http://www.e-crisp.org/E-CRISPR/), Zifit CRISPR (http://zifit.partners.org/ZiFiT/) (ZingFinger Consortium), Cas9des : //Cas9.cbi.pku.edu.cn/) (Beijing University), CRISPRdirect (http://crispr.dbcls.jp/) (Tokyo University), CRISPR-P (http://cbi.hzau.edu) .Cn / crispr /) (China Agricultural University), Guide RNA Target Design Tool. It can be determined by using (https://wwws.blueheronbio.com/external/tools/gRNASrc.jsp) (BlueHeronBiotech) and the like.

Also, depending on the CRISPR / Cas system, the guide RNA may not contain tracrRNA. Examples of the CRISPR / Cas system using such a guide RNA as a component include the CRISPR / Cpf1 system.

A person skilled in the art can prepare a guide RNA according to the present invention using an in vitro transcription reaction system (for example, T7 DNA polymerase system) using the DNA encoding the same as a template based on the sequence information.

Further, the expression vector containing the polynucleotide encoding the guide RNA according to the present invention can be appropriately prepared by a person skilled in the art as in the fusion protein described above, but the promoter contained in the expression vector has a short code. From the viewpoint of efficient transcription of the region, a polIII promoter (for example, U6, H1, SNR6, SNR52, SCR1, RPR1 promoter) is preferably used.

The DNA editing system of the present invention may constitute a kit containing a combination of the fusion protein of the present invention and a guide RNA or the like. The kit may further include one or more additional reagents. Examples of such additional reagents include, but are not limited to, dilution buffers, reconstruction buffers, wash buffers, nucleic acid transfer reagents, protein transfer reagents, and control reagents. The kit may also include additional instructions for carrying out the methods of the invention.

(How to make cells with edited DNA)
The present invention provides a method for producing a cell in which DNA has been edited, which comprises introducing the above-mentioned genome editing system into the cell.

The cells used in the present invention may be eukaryotic cells such as animal cells, plant cells, algae cells and fungal cells, or prokaryotic cells such as bacteria and paleontology. Eukaryotic cells are preferred.

Examples of "eukaryotic cells" include animal cells, plant cells, algae cells, and fungal cells. In addition to mammalian cells, animal cells include cells of fish, birds, reptiles, amphibians, and insects.

The "animal cell" includes, for example, cells constituting an individual animal, cells constituting an organ / tissue extracted from an animal, cultured cells derived from an animal tissue, and the like. Specifically, for example, germ cells such as egg mother cells and sperm; embryo cells of embryos at each stage (for example, 1-cell stage embryo, 2-cell stage embryo, 4-cell stage embryo, 8-cell stage embryo, 16-cell stage). Embryos, mulberry stage embryos, etc.); Stem cells such as induced pluripotent stem (iPS) cells and embryonic stem (ES) cells; fibroblasts, hematopoietic cells, neurons, muscle cells, bone cells, hepatocytes, pancreatic cells , Somatic cells such as brain cells and kidney cells. As the oocyte used for producing the genome-editing animal, pre-fertilization and post-fertilization oocytes can be used, but post-fertilization oocytes, that is, fertilized eggs are preferable. Particularly preferably, the fertilized egg is of a pronuclear stage embryo.

"Mammals" is a concept that includes humans and non-human mammals. Examples of non-human mammals include ferrets such as cows, wild boars, pigs, sheep and goats, cloven-hoofed animals such as horses, rodents such as mice, rats, guinea pigs, hamsters and squirrels, and lagomorphs such as rabbits. , Dogs, cats, ferrets and other meats. The non-human mammal may be a domestic animal or a companion animal (pet animal) or a wild animal.

Examples of "plant cells" include cells of cereals, oil crops, forage crops, fruits, and vegetables. The "plant cell" includes, for example, cells constituting an individual plant, cells constituting an organ or tissue separated from a plant, cultured cells derived from a plant tissue, and the like. Examples of plant organs and tissues include leaves, stems, shoot apex (growth point), roots, tubers, callus and the like. Examples of plants include rice, corn, banana, peanut, sunflower, tomato, rape, tobacco, wheat, barley, potato, soybean, cotton, carnation, etc., and their breeding materials (eg, seeds, tubers, tubers, etc.) ) Is also included.

Examples of methods for introducing a genome editing system into cells include electroporation, calcium phosphate method, liposome method, DEAE dextran method, microinjection method, cationic lipid-mediated transfection, electroporation, transduction, and Will vector. Examples include methods such as infection. Such a method is described in many standard laboratory manuals such as "Leonard G. Daviset al., Basic methods in molecular biology, New York: Elsevier, 1986".

The fact that the desired DNA editing occurs in the cells produced in this manner can be determined by those skilled in the art using known DNA analysis methods (for example, PCR method, sequencing method, Southern blotting method). Can be confirmed.

In addition, by comprehensively determining the sequence by using next-generation sequencing (NGS; Next Generation Sequencing) or single molecule sequencing method, not only desired DNA editing but also the presence or absence of off-target can be confirmed. can do. The next-generation sequencing method is not particularly limited, but is a synthetic sequencing method (sequencing-by-sequencing, for example, Sequencing by Illumina Solexa Genome Analyzer, Hiseq®, Nextseq, Miseq or Miniseq), Pyro. Sequencing method (for example, sequencing by sequencer GSLX or FLX manufactured by Roche Diagnostics (454) (so-called 454 sequencing)), rigase reaction sequencing method (for example, SoliD® manufactured by Life Technology Co., Ltd.) Alternatively, sequencing with 5500 xl) can be mentioned. Examples of the single molecule sequencing method include PacBio RS II or PacBioSequel system manufactured by Pacific Biosciences of California, PromethION, Gilead ION or MinION manufactured by Oxford Nanopore Technologies.

Hereinafter, the present invention will be described in more detail based on Examples, but the present invention is not limited to the following Examples. The materials and methods used in the examples will be described below.

(Preparation of base editor expression plasmid)
All base editor expression plasmids were prepared using the same backbone sequence used in pCMV-BE3 (Catalog No. 73921, manufactured by Addgene).

The BE4 plasmid (pCMV-BE4), BE4max plasmid (pCMV-BE4max), ABE7.10 plasmid (pCMV-ABE7.10) and ABEmax plasmid (pCMV-ABEmax) having the same backbone were obtained from Addgene (each). Catalog numbers: 100802, 112093, 102919 and 112095).

The other single-function base editor and dual-function base editor were constructed by concatenating PCR fragments with Gibson assembly as follows.

The Target-AID plasmid (pCMV-Target-AID) is a backbone fragment in which two fragments encoding the N-terminal and C-terminal half of Target-AID are amplified from pCMV-ABE 7.10 using the primer pair RS047 / RS8. Built by assembling. The two fragments were amplified from pcDNA3.1_pCMV-nCas-PmCDA1-ugi pH1-gRNA (HPRT) (Catalog No. 79620, manufactured by Addgene) using primer pairs RS045 / HM129 and HM128 / RS046, respectively.

In order to construct the Target-AIDmax plasmid (pCMV-Target-AIDmax), the pUC-optimized-PmCDA1-ugi plasmid, which encodes the C-terminal region of Target-AIDmax with optimized codons, was first subjected to a gene synthesis service of GenScript. Used and built. This C-terminal fragment is amplified by the primer pair SI1304 / SI1307 and amplified from pCMV-BE4max using the primer pair SI945 / SI1308, and the backbone amplified from pCMV-ABEmax using SI1310 / SI1309. Concatenated with the fragment.

The BE4max (C) plasmid (pCMV-BE4max (C)) was constructed by replacing the C-terminal region of Target-AIDmax with two codon-optimized rAPOBEC1 and BE4max UGI domains. To this end, the nCas9 fragment obtained from pCMV-Target-AIDmax using SI447 / SI1105, the rAPOBEC1 and two UGI fragments obtained from BE4max using the primer pairs SI1352 / SI1357 and SI1359 / SI1350, respectively. Was ligated to the backbone fragment obtained from pCMV-BE4max using the primer pair SI1351 / SI448.

The Target-ACE plasmid (pCMV-Target-ACE) contains a fragment encoding the plasmid skeleton, ABE7.10 amplified from pCMV-ABE7.10 using the primer pair RS047 / RS052, and the primer pair RS05RS / RS04. It was constructed by ligating a fragment encoding the C-terminal region of Target-AID amplified from pcDNA-pCMV-nCas9 using.

The Target-ACEmax plasmid (pCMV-Target-ACEmax) was obtained from the ABEmax fragment obtained from pCMV-ABEmax using the primer pair SI945 / SI1305 and from the pUC optimized PmCDA1-ugi using the primer pair SI1304 / SI1307. It was constructed by ligating a fragment encoding the C-terminal region of -AIDmax and a fragment encoding the plasmid skeleton obtained from pCMV-ABEmax using the primer pair SI1310 / SI1309.

The ACBEmax plasmid (pCMV-ACBEmax) is an rAPOBEC1 domain, 2xUGI, prepared to construct pCMV-BE4max (C) with the ABEmax fragment obtained from pCMV-Target ACEmax using the primer pair SI447 / SI1105. The domain and the three fragments encoding the two backbone fragments were constructed by concatenation. Note that the two UGI domains have non-synonymous nucleotide substitutions in the GS linker between tandem UGIs (SGGSG [G> E] SGGS).

(Preparation of gRNA expression plasmid)
The gRNA spacer insert was prepared by a single pot reaction on phosphorylated and annealed ssDNA pairs. To prepare each spacer fragment, a T4 polynucleotide kinase reaction sample was prepared using two ssDNAs according to the manufacturer's (Takara) protocol and placed in a thermal cycler at 37 ° C. for 30 minutes and 95 ° C. for 5 minutes. Then, starting from 95 ° C., the reaction was carried out for 70 cycles (12 seconds for each cycle) while lowering the temperature by 1 ° C. for each cycle, and then maintained at 25 ° C.

The annealed spacer insert was then ligated to the pU6-gRNA cloning backbone (pSI-356) by golden gate assembly using BsmBI (NEB) and T4 DNA ligase (NEB). This assembly was performed with a thermal cycler under the following conditions.
15 cycles of 37 ° C. for 5 minutes, 20 ° C. for 5 minutes and 55 ° C. for 30 minutes, then maintained at 4 ° C.

(Preparation of wrench virus base editing reporter plasmid)
The C → T reporter system was designed so that when C → T base editing occurs in the antisense strand, GTG changes to ATG and the start codon is restored, as shown in FIG. 3A. As shown in FIG. 3B, the A → G reporter system translates downstream EGFP by changing TAA to CAA and disrupting the stop codon when base editing of A → G occurs in the antisense strand. Designed to progress.

In order to construct the lentivirus C → T base editing reporter plasmid (pLV-SI-112) and its positive control plasmid (pRS112), the reporter cassette fragment was added to the primer set 112-V4-BC2-FW / SI680 and RS204 /. Using SI666, pLV-eGFP (catalog number: 36083 manufactured by Addgene) was used as a template for amplification. Each of them was inserted into the EcoRI and BamHI recognition sites of the pLVSIN-CMV-Puro backbone vector (manufactured by Takara) using T4 DNA ligase (manufactured by NEB) and cloned.

For the lentivirus A → G base editing reporter plasmid (pLV-SI-121) and its positive control plasmid (pLV-SI-122), each reporter cassette fragment was used, and the primer sets SI760 / SI680 and SI761 / SI680 were used, respectively. It was constructed in the same manner as the lentivirus C → T base editing reporter plasmid except that it was amplified.

The sequences of the various plasmids prepared as described above were confirmed by sanger sequencing.

(Selection of gRNA target site)
To select gRNA target sites with cytosine and adenine for ampli-consequencing assays, first polycytosine repeats (poly-C), polyadenine repeats (poly-A), alternate polyadenine / cytosine repeats (poly) Target sites with -AC) and alternating polycytosine / adenine repeats (poly-CA) were searched in the human genome (hg19). The poly-C target site was conditioned to have cytosine that fills a 7-bp sliding window in which the 5'end shifts from PAM to the range of -24 bp to -16 bp at 2 bp intervals. The poly-A target site was conditioned to have adenine at the end of 5'filling a 6-bp sliding window that shifts from PAM to the range of -21 bp to -13 bp at 2 bp intervals. The poly-AC target site and the poly-CA target site are subject to each repetitive sequence satisfying a 6-bp sliding window in which the 5'end shifts from PAM to the range of -24 bp to -14 bp at 2 bp intervals. did. In addition, candidate target sites containing 4 bp or more of homopolymers in the gRNA seed region ranging from -8 bp to -1 bp and overlapping the annotated exons were excluded. For the positions of the poly-C and poly-A sliding windows, the two candidate sites with the highest predicted gRNA activity scores were selected, respectively. Similarly, one candidate site was selected for each of the poly-AC and poly-CA slide window positions.

By using amplicon sequencing primers designed for these target sites and strongly amplifying by the amplification sequencing library preparation protocol by the present inventors, 7 poly-C, 7 poly-A, Six poly-AC and four poly-CA target sites were further screened. The selected target sites of poly-C and poly-A contain both cytosine and adenine bases. In addition to these 24 target sites, 24 target sites were screened from the gRNA library collection previously prepared by us for another assay. Each of these target sites should contain one or more cytosines and one or more adenines in the -20 bp to -14 bp region. A total of 47 on-target sites were analyzed because EGFP control data could not be obtained in the CUL3-NGG site 2 amplicon sequencing assay.

In off-target site analysis by ampli-sequencing, two on-target sites of FANCF and one on-target site of the EMX1 gene having both cytosine and adenine bases in the region ranging from -20 bp to -14 bp, and previous reports. Identified by GUIDE-seq in (Tsai, S.Q. et al. Nat Biotechnol 33, 187-197 (2015). And Kleinstiber, BP et al. Nature 529, 490-495 (2016).). We have selected many off-target sites. Five of the previously reported off-target sites were selected for each on-target site, but one of the EMX1 off-target sites was excluded from the analysis because it was not amplified by the amplicon sequencing library preparation protocol. did.

(Cell culture)
HEK293Ta cells were purchased from GeneCopoea and supplemented with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific) and 1% penicillin-streptomycin (Sigma) in modified Dulbecco Eagle's Medium (DMEM, Sigma). Was maintained at 37 ° C. under 5% CO ₂ conditions. In addition, cells were routinely checked for mycoplasma contamination by nested PCR using a culture medium as a template.

(Cell line with base editing reporter system)
A C → T and A → G reporter system and a positive control system containing the expected mutations in each were introduced into human embryonic kidney cells HEK293Ta cells by lentivirus transmission.

~ 2 × 10 ⁵ cells / well were seeded on 6-well plates 1 day prior to transfection for wrench virus packaging. In each packaging reaction, 489 ng of lentivirus plasmid, two helper plasmids, psPAX2 (manufactured by Addgene, catalog number: 12260), pMD2. G (Addgene, Catalog No .: 12259) 366 ng and 122 ng, respectively, in 1 mg / mL polyethyleneimine MAX (PEI, Polysciences) 9.38 μL and 300 μL phosphate buffered saline (PBS). I did it. The next day, the culture medium was changed to fresh medium, and two days later, the culture supernatant containing lentivirus particles was collected and dispensed into a 1.5 mL tube. The resulting virus sample was stored at −80 ° C. until infection.

In Lentivirus infection, seeded in 6-well plates with 2 mL DMEM to ~ 2 × 10 ⁵ cells / well and incubated for 24 hours. The virus supernatant was thawed at room temperature, 1 μL of 8 mg / mL polybrene (manufactured by Sigma) was mixed and added to each cell sample. The day after infection, ~ 5 × 10 ³ infected cells were reseeded into 96-well culture plates to measure their functional titers by the CellTiter-Glo assay (Promega). Two days after infection, 2.0 μg / mL puromycin (manufactured by Thermo Fisher Scientific) was added to the medium, followed by incubation for 3 days to select normally transgenic cells. After selection of puromycin, it was confirmed that background fluorescence was not generated and EGFP was expressed in the reporter cell line and the corresponding positive control cell line, respectively.

(Transfection during EGFP reporter activation assay)
The cells into which the base editing reporter system was introduced and the corresponding control cells were placed in a 24-well plate (manufactured by AS ONE) coated with collagen I of 500 μL DMEM so as to have a density of ~ 5 × 10 ⁴ cells / well. Sown. The next day, 1 mg / mL PEI 1.2 μL, 50 μL PBS, base editor expression plasmid 300 ng, and 100 ng gRNA expression plasmid were mixed and then incubated for 20 minutes at room temperature before being added to each well for transfection.

In the experiment in which each base editor was mixed and used, two kinds of base editor expression plasmids were mixed at a mass ratio of 1: 1 and 300 ng was used. Fluorescence imaging was performed 3 days after transfection using a confocal microscope InCellAnalyzer6000 (manufactured by GE Healthcare) equipped with a 20x objective lens. The experiment was repeated 4 times. At one time, the cell nuclei were stained with 10 mg / mL Hoechst 33342 (manufactured by Thermo Fisher Scientific).

(Transfection during genome on-target and off-target assays)
HEK293Ta cells were seeded in a collagen I coated 96-well plate (manufactured by Asone) with 200 μL DMEM to a density of ~ 5 × 10 ³ cells / well. The next day, 0.48 μL PEI, 50 μL 1 × PBS, 120 ng base editor expression plasmid or control EGFP expression plasmid (pLV-eGFP), and 40 ng gRNA expression plasmid were mixed and then at room temperature before being added to each transfection well. Incubated for 15 minutes. This experiment was independently repeated 3 times.

(Transfection during genome-wide DNA and RNA off-target assay)
HEK293Ta cells were seeded on a collagen I coated 6-well plate (manufactured by Asone) with 2 mL DMEM to a density of ~ 2 × 10 ⁵ cells / well. The next day, 3.0 μL of 1 mg / mL PEI, 200 μL 1 × PBS, 666 ng base editor expression plasmid or control EGFP expression plasmid (pLV-eGFP), and 333 ng EMX1-targeting gRNA plasmid were mixed and incubated at room temperature for 15 minutes. , Added to each well. This experiment was independently repeated twice.

(Amplicon sequencing in EGFP reporter activation assay)
After confocal imaging, the culture medium was aspirated and 200 μL of freshly prepared 50 mM NaOH was added to each cell sample in a 24-well plate. Then, 100 μL of it was transferred to a 96-well PCR plate (manufactured by Nippon Genetics), heated at 95 ° C. for 15 minutes, cooled to 4 ° C., and then 20 μL of 1M Tris-HCl (pH 8.0) was added. Neutralized. Using the cell lysate thus prepared as a template, the corresponding 1st HTS primer was used to amplify the target region in each sine pull.

PCR is performed in a 20 μL solution containing 1 μL template, 10 μM primers, 1 μL, 0.2 μL Phusion DNA polymerase, 5 × Phusion HF buffer (manufactured by NEB) and 1.6 μL of 2.5 mM dNTPs under the following temperature cycle conditions. It was done in.
The cycle of 98 ° C. for 30 seconds, then 98 ° C. for 10 seconds, 60 ° C. for 10 seconds and 72 ° C. for 10 seconds was repeated 30 times, and for the final extension reaction, 72 ° C. for 5 minutes.

The PCR product thus obtained was subjected to electrophoresis on a 2% agarose gel. Further, 1 μL of a 10-fold diluted solution was used as a template, and the mixture was re-amplified in a 20 μL solution containing a custom Illumina index primer under the following temperature cycle conditions.
The cycle of 98 ° C. for 30 seconds, then 98 ° C. for 10 seconds, 60 ° C. for 10 seconds and 72 ° C. for 30 seconds was repeated 15 times, and for the final elongation reaction, 72 ° C. for 5 minutes.

Each index library thus obtained was subjected to electrophoresis on a 2% agarose gel, and a band having an expected molecular weight was recovered using a FastGene Gel / PCR extraction kit (manufactured by Nippon Genetics).

(Amplicon sequencing in genome on-target and off-target assays)
Three days after transfection, the culture was removed and 50 μL of freshly prepared 50 mM NaOH was added to each cell sample in a 96-well plate. The cell sample was transferred to a 96-well qPCR plate (BioRad), sealed with an optically transparent adhesive PCR seal (BioRad), centrifuged at 2,400 rpm for 2 minutes, heated at 95 ° C. for 15 minutes, 4 The mixture was cooled to ° C., followed by the addition of 5 μL of 1M Tris-HCl (pH 8.0) for neutralization. The plate containing the cell lysate was centrifuged again and stored at −20 ° C. Each target region was amplified using the corresponding 1st HTS primer pair.

PCR is performed in a 20 μL solution containing a 2 μL genomic DNA template, 8.3 μM each primer 1.20 μL, 0.2 μL Phaseion DNA polymerase, 5 × Phaseion HF buffer, and 1.6 μL 2.5 mM dNTPs in the following temperature cycle. It was done under the conditions.
First, a cycle of 98 ° C. for 30 seconds, 98 ° C. for 10 seconds, 60 ° C. for 10 seconds and 72 ° C. for 60 seconds was repeated 30 times, and for the final extension reaction, 72 ° C. for 5 minutes.

Each time a reproduction experiment was performed, 3 μL of each PCR product of the same base editor reagent was pooled and purified using 1.8 × volume of Agencourt AMPure XP magnetic beads (manufactured by Beckman Coulter).

Using 1 μL of the first PCR product of 10 ng / μL as a template and using a custom Illumina index primer, a PCR reaction was carried out in a 20 μL solution under the following temperature cycle conditions to amplify the purified product.
First, a cycle of 98 ° C. for 30 seconds, then 98 ° C. for 10 seconds, 65 ° C. for 10 seconds, and 72 ° C. for 90 seconds was repeated for 15 cycles, and for the final extension reaction, 72 ° C. for 5 minutes.

Each index library thus obtained was subjected to electrophoresis on a 2% agarose gel, and a band having an expected molecular weight was recovered using a FastGene Gel / PCR extraction kit (manufactured by Nippon Genetics).

The sequence library was quantified by qPCR using the KAPA library quantification kit Illumina (manufactured by KAPA Biosystems) for multiplexing. The multiplexed library uses Illumina HiSeq2500 (TruSe rapid SBS kit; 2 × 151 bp paired-end) or MiSeq (MiSeq v3 kit; 2 × 2200 bp paired-end) with 20% to 30% of PhX. As a control, it was quantified by the same qPCR protocol.

(Preparation of RNA-seq and WES library)
Three days after transfection, cells were lysed with trypsin and dispensed into two 1.5 mL tubes for transcriptome sequencing (RNA-seq) and total exome sequencing (WES).

To prepare the RNA-seq library, the cells were subjected to centrifugation at 1,000 rpm for 5 minutes, and the culture supernatant was removed. Then, total RNA was extracted using ISOSPIN Cell & Tissue RNA (Nippon Gene), and an RNA-seq library was prepared using a TRUSeq Stranded mRNA library preparation kit (manufactured by Illumina).

For the preparation of the WES library, genomic DNA was extracted using NucleoSpin Tissue (Machery Nagal). Then, 500 ng of genomic DNA was fragmented into an average size of 150 to 300 bp in a 50 μL solution using an ultrasonic irradiation device (manufactured by Waken Yakuhin Co., Ltd., product name: Covalis E-220). Then, end repair, A-tailing and SureSelect adapter ligation (manufactured by Agilent) were performed.

The adapter-binding DNA was concentrated using the KAPA HyperPrep kit (KAPA Biosystems). Then, 750 ng of the pre-amplified DNA was hybridized to the SureSelectXT Human All Exon V3 kit probe (manufactured by Agilent) over 20 hours.

Post-capture DNA library amplification was performed using KAPA DNA polymerase and SureSelect Indexing post-capture polymerase chain reaction primers to index the library. Finally, the library was purified using Agecurt AMPure XP beads. Fragment size distributions and yields of RNA-seq and WES libraries were quantified using a LabChip GX electrophoresis system (manufactured by PerkinElmer). After multiplexing, the final library was sequenced with Illumina NovaSeq 6000 (S2 flow cell; 2 x 101 bp paired end).

(Amplicon sequencing analysis)
First, in order to identify the read of the custom sample index and the demultiplexed pair end, a common adapter sequence is used with the Blastn-short option, and NCBI BLAST + (version 2.7.0) is used for the amplicon sequencing read. Mapped to. Next, to generate a further mapped, merged sequencing read to the corresponding reference sequence of the target region using the EMBOSS needle package (version 6.6.0) with an ID threshold of 80%. The paired end reads of each sample were merged using FLASH (version 1.2.0).

In single and multiple edited spectrum analyzes using genome-on-target and off-target amplicon sequencing data, the EGFP transfection control data always elicited the same background signal.

(RNA and DNA mutant detection pipeline)
RNA-seq and WES base call files were multiplexed using bcl2fastq2 (version v2.20.0). To eliminate sequencing coverage bias for variant calls, all RNA-seq and WES libraries use sextk (version 1.3-r107-dirty) with 74 million reads and 94 million reads per sample, respectively. Was randomly subsampled. The subsampled RNA-seq reads were then referenced using STAR (version 2.7.3a) and transcript annotation GTF (GENCODE Release 22 GRCh38.p2) to reference the human genome GRCh38. d1. It was mapped to vd1 and deduplicated using Picard MarkDuplicates (version 2.0.1). Subsampled DNA reads were also processed for mapping according to the National Cancer Center Genome Data Commons DNA-Seq analysis pipeline. That is, the read is referenced by the BWA-MEM (version 0.7.15) Human Genome GRCh38. d1. Aligned with vd1, PCR deplicates were removed using Picard Mark Duplicates (version 2.0.1). The GATK Haplotype Caller (version 4.1.4.1) is available at the Reference Human Genome GRCh38. d1. It was used to detect both DNA and RNA variants of vd1. Mutation positions detected by the GATK haplotype caller are described in Grunewald, J. et al. et al. Further filtering as described in Nat Biotechnol 37, 1041-1048 (2019) (estimation of DNA off-target risk)
To investigate the DNA off-target effect on various base editing methods caused by non-specific binding of the target gRNA, we selected three gRNAs whose off-target sites target the 8EMX1 and FANCF genes commonly observed by GUIDE-seq. It was used (Tsai, S.Q. et al. Nat Biotechnol 33, 187-197 (2015).). After transfecting HEK293Ta cells with a base editor and on-target gRNA reagent, amplifiers at 5 off-target sites for FANCF target site 1, 5 off-target sites for FANCF target site 2, and 4 off-target sites for EMX1 target site. Base editing efficiency was analyzed by reconsequencing. At all tested off-target sites, the median relative off-target activity normalized to the corresponding on-target activity was calculated as the DNA off-targeting risk score for each base editing method.

(Base editing prediction model)
Model Training Conditional probabilistic models were trained using amplicon sequencing datasets for various target sites. Observed edits with a relative read frequency of less than 1 × 10 ^-4 were first removed to minimize the effects of potential sequence errors during the training process. The nucleotide base rearrangement state at the ibp position from PAM at the target site was defined as s_i, and the probability was defined as P (Si). For each target region, P (Si) and P (Sj│Si) were calculated for each combination of i and j in the designated region (i ≠ j). The training model was finally constructed by the mean values of P (Si) and P (Sj│Si) over the various training target sites observed in 100 or more leads.

Prediction of the result of base editing The base editing pattern in the window from PAM to m bp to n bp is Sm, n, and the dislocation-state character strings sm, sm + 1, ..., Sn-1, sn. Using a training model that can be represented by, the frequency of predetermined results Sm, n at the test site was predicted by the following formula.

Basically, in order to predict the frequency of a particular edit result with multiple base translocations, this prediction model takes into account other independent base translocation patterns and of the base translocations at all edited positions. Calculate the geometric average of probabilities. The specific P (s_j | s_i) lacking training data is ignored (treated as 1), and the P (s_i) lacking training data is 0.

Verification of Base Editing Prediction Model The base editing prediction model was evaluated by 5-fold, 15-fold, and leave-one-out cross-validation experiments. For the target site tested, the frequency of all base edit result patterns detected in the ampli-consequencing dataset for windows ranging from -25 bp to -5 bp from PAM does not overlap with the test target site amplifiers at other target sites. Predicted by training reconsequencing data. In the k-fold cross-validation experiment, [47 / k] target sites were randomly selected as test samples from 47 target sites. The edited results were predicted by training the amplicon sequencing data of the remaining target sites, and the test sample was randomly changed and repeated 100 times. For leaf-one-out cross-validation, amplicon sequencing data from 46 other target regions were used to predict the editing results for each target site. For prediction performance, first, the predicted editing frequency is converted into a relative editing frequency among all edited leads, and in the case of multiple, one prediction result is randomly selected for each editing result pattern, and prediction and experimentation are performed. It was measured by calculating Pearson's correlation coefficient between measurements.

(Simulation of co-editing frequency in synthesized target sequences)
100 synthetics consisting only of cytosine and / or adenine bases in the region from -20 bp to -1 bp with respect to PAM to predict multidimensional co-edited spectra for various base editing methods using base editing prediction models. The target sequence was simulated. For each target region, C → T and A → G editing may all occur in with the results (of a total of 2 ²⁰ results), using the training bases edited prediction model from amplicon sequencing data every 47, predicted .. Meaning of homologous and heterogeneous dinucleotide edited spectra was calculated using all predicted frequencies. Trinucleotide editing spectra were also predicted for simulated sequences in which poly-AC extends from -20 bp to -1 bp with respect to PAM.

(Codon conversion matrix and bystander mutation risk)
To estimate the codon conversion potential and the risk of bystander mutations in each base editing method, the codon conversion matrix (CCM) with or without bystander mutations that generate unwanted mutations with the target codon conversion. Was generated.

First, for 11,250,496 kinds of source codons (hg38) in the human genome, sites that could be targets of gRNA were screened in the region ± 25 bp from the target codon. For all gRNAs, the probability for all C → T and A → G editing patterns that can occur in the ± 15bp region of the target codon triplet is a base editing prediction model trained by amplicon sequencing data for all 47 genomic sites. Was predicted using. Then, the possibility of mutation from the target source codon to each destination codon without bystander mutation is defined as the maximum probability of producing the desired result among all possible gRNAs. did. If bystander mutations can occur, the predicted probabilities for the outcome of base editing with target source codon conversion are all summed per gRNA, and the maximum integration probability among the possible gRNAs is defined as the conversion probability. did. The conversion probability for codons that do not have a potential gRNA was defined as 0 for any destination codon type. After calculating the conversion probabilities to various destination codons for all genomic codons, a CCM was finally generated to show the frequency of each source destination codon conversion type with a conversion probability threshold of 5%. For various source-destination codon types, bystander mutation risk scores were calculated by dividing the CCM frequency that allows bystander mutations by those that do not tolerate bystander mutations.

(ClinVar analysis)
For the pathological C ・ G → TA and A ・ T → G ・ C SNVs reported in the ClinVar database, gRNA target sites that may correct each mutation were determined within the ± 25 bp region from the target mutation. First screened. Using a base-edited predictive model trained with all amplicon sequencing datasets, the probability that mutations will be modified by various gRNA target sites in the ± 15 bp region from the target mutation without inducing unwanted bystander mutations. Predicted. For each mutation, its corrected probability was defined as the maximum probability of target codon conversion among those induced by various gRNAs. Finally, to estimate the global potential of a dual-function base editor to modify two heterogeneous disease mutations simultaneously, two heterogeneous suddens both predicted to be correctable with a correction probability threshold of 5%. The combination of mutations was counted. To reduce computational costs, we limited the heterologous mutation space combinations to those within the same gene.

(Statistical analysis)
All genome editing experiments were repeated at least 3 times with independent cell culture samples. Statistical tests were performed with Python (version 3.7.4) or R (version 3.6.0.) And Scipy (version 1.4.1).

The examples carried out using the above-mentioned materials and methods will be described below.

Cytidine deaminase and adenosine deaminase in a single nCas9 (D10A) with the aim of providing a base editing tool that allows simultaneous base substitution of both C → T and A → G. Three types of base editors, Target-ACE, Target-ACEmax, and ACBEmax, which have both functions by fusion, were developed and tested (Fig. 1).

In Target-ACE, nCas9 fused to PmCDA1 derived from Target-AID and TadA heterodimer derived from ABE-7.10 are fused as C-terminal and N-terminal, respectively, for consistent single-function base editing. It is configured with other functional domains that exist. The amino acid sequence of Target-ACE and the nucleotide sequence encoding are shown in SEQ ID NOs: 4 and 3, respectively. In the amino acid sequence shown in SEQ ID NO: 4, positions 2 to 167 indicate a modified TadA (TadA *), positions 168 to 171 indicate a GGS linker, and positions 172 to 175 indicate a GGS linker, and positions 176 to 175. Positions 191 indicate the XTEN linker, positions 192 to 195 indicate the GGS linker, positions 196 to 199 indicate the GGS linker, positions 200 to 365 indicate TadA, and positions 366 to 369 indicate the GGS linker, and positions 370 to 373. Positions indicate GGS linkers, positions 374 to 389 indicate XTEN linkers, positions 390 to 393 indicate GGS linkers, positions 394 to 397 indicate GGS linkers, and positions 398 to 1764 indicate SpCas9 (D10A), 1765. Positions ~ 1774 indicate 2 × GS linker, positions 1775-1831 indicate mutant SH3, positions 1832 to 1857 indicate 3 × FLAG, positions 1858 to 2065 indicate PmCDA1, and positions 2066 to 2073 indicate mononode type nuclei. A translocation signal (nuclear localization signal derived from SV40 T antigen) is shown, and positions 2077 to 2159 indicate a uracil glycosilase inhibitor.

Target-ACEmax was constructed by applying GenScript codon optimization and addition of a bifurcated nuclear localization signal (NLS) to the N-terminal to the Target-ACE (Fig. 2). The amino acid sequence of Target-ACEmax and the encoding nucleotide sequence are shown in SEQ ID NOs: 2 and 1, respectively. In the amino acid sequence shown in SEQ ID NO: 2, positions 2 to 19 indicate a binode-type nuclear localization signal, positions 20 to 185 indicate a modified TadA (TadA *), and positions 186 to 189 indicate a GGS linker. Positions 190 to 193 indicate the GGS linker, positions 194 to 209 indicate the XTEN linker, positions 210 to 213 indicate the GGS linker, positions 214 to 217 indicate the GGS linker, and positions 218 to 383 indicate TadA. , 384-387 positions indicate GGS linker, 388-391 positions indicate GGS linker, 392-407 positions indicate XTEN linker, 408-411 positions indicate GGS linker, and 412-415 positions indicate GGS linker. Positions 416 to 1782 indicate SpCas9 (D10A), positions 1783 to 1792 indicate 2 × GS linker, positions 1793 to 1849 indicate mutant SH3, positions 1850 to 1875 indicate 3 × FLAG, and positions 1876 to 2083. Indicates PmCDA1, positions 2084 to 2091 indicate a mononode-type nuclear localization signal (nuclear localization signal derived from SV40 T antigen), and positions 2095 to 2177 indicate a uracil glycosylase inhibitor.

ACBEmax was constructed by replacing the codon-optimized PmCDA1 domain in Target-ACEmax with the codon-optimized cytidine deaminase domain rAPOBEC1 derived from BE4max.

The amino acid sequence of ACBEmax and the encoding nucleotide sequence are shown in SEQ ID NOs: 6 and 5, respectively. In the amino acid sequence shown in SEQ ID NO: 6, positions 2 to 19 indicate a binode-type nuclear localization signal, positions 20 to 185 indicate a modified TadA (TadA *), and positions 186 to 189 indicate a GGS linker. The 190-193 positions indicate the GGS linker, the 194-209 positions indicate the XTEN linker, the 210-213 positions indicate the GGS linker, the 214-217 positions indicate the GGS linker, and the 218-383 positions indicate the TadA. , 384-387 positions indicate the GGS linker, 388-391 positions indicate the GGS linker, 392-407 positions indicate the XTEN linker, 408-411 positions indicate the GGS linker, and 412-415 positions indicate the GGS linker. Positions 416 to 1782 indicate SpCas9 (D10A), positions 1783 to 1792 indicate a 2 × GS linker, positions 1793 to 1849 indicate mutant SH3, positions 1850 to 1875 indicate 3 × FLAG, and positions 1876 to 2103. Indicates APOBEC1, positions 2104 to 2186 indicate a uracil glycosylase inhibitor, positions 2187 to 2196 indicate a 3 × GGS linker, positions 2197 to 2279 indicate a uracil glycosylase inhibitor, and positions 2280 to 2283 indicate a GGS linker. , 2284-2300 indicate a bifurcated nuclear localization signal.

Also. As controls for these dual-function base editors, in addition to Target-AID, BE4max, ABE, and ABEmax, codon-optimized Target-AIDmax and BE4max (C) were constructed. BE4max (C) is obtained by replacing the C-terminal PmCDA1 domain derived from Target-AIDmax with the rAPOBEC1 domain derived from BE4max.

To test the base editing activity of single-function and dual-function base editors in living cells, we have designed the corresponding base substitutions to activate the expression of the EGFP protein, C → T. And A → G base editing reporter cells were constructed (FIGS. 3A and 3B). All monofunctional CBEs could only activate C → T reporter cells. In addition, ABEs could only activate A → G reporter cells (Fig. 4). On the other hand, three types of dual-function base editors, Target-ACE, Target-ACEmax and ACBEmax, control them (a mixture of enzymes corresponding to each dual-function base editor; Target-AID + ABE, Target-AIDmax + ABEmax, BE4max (C). ) + ABEmax, and BE4max + ABEmax), both C → T reporter cells and A → G reporter cells were activated. These results were also confirmed by amplicon sequencing of gRNA target regions in reporter cells (FIGS. 5A-D).

In order to characterize the C → T and A → G base editing activities in various base editing methods, the base editing spectra at 47 genomic target sites of human embryonic kidney (HEK293Ta) cells were subjected to triple amplicon sequencing (1, 833 assay). As a result, by taking the average editing frequency of C → T or A → G in cytosine or adenine at each site related to PAM, the dual-function base editor has similar base editing properties from the corresponding single-function base editor. It became clear that it inherited (Fig. 6). In addition, the C → T and A → G edit frequencies at 47 endogenous target sites containing different numbers of cytosine and adenine were significantly different, but were generally consistent with the base edit spectral data (FIG. 7).

The amplicon sequencing data for 47 genomic target sites involves editing both C → T and A → G by a dual-function base editor and four controls (four combinations of corresponding base editors). An edit result pattern of 722 was observed with a read frequency of 0.1% or more (FIG. 8). Overall, it was revealed that Target-ACEmax and Target-ACE, and their corresponding base editor mixtures, exhibit clusters of multi-base editing patterns that differ from ACBEmax and its corresponding base editor mixtures. To observe co-editing patterns at 47 different target sites, dinucleotide homologous co-editing spectra and dinucleotide heterogeneous co-editing spectra were then examined for each of the 13 base editors. The mean co-editing frequency for each cytosine-cytosine pair, adenine-adenine pair or cytosine-adenine pair at different locations from the PAM was calculated using the amplicon sequencing results of the relevant target sites. In the various base editor reagents, the plurality of C → T edited spectra and the plurality of A → G edited spectra were reacquired with the same tendency as the average frequency of C → T and A → G at 47 target sites. When editing C → T and A → G at the same time, Target-ACEmax and Target-AIDmax + ABEmax were 19.2% and 21. Around cytosine located at -18 bp and adenine located at 15 bp with respect to PAM, respectively. Efficient co-editing spectra were shown at a peak height of 0% (FIGS. 9 and 10). ACBEmax and BE4max + ABEmax showed different efficient co-editing spectra, with peak frequencies in the cytosine-adenine position combination at -12 and -15bp being the highest of all base editor reagents (23.2% each). And 23.6%).

To investigate the DNA off-target effects of various base editing methods caused by non-specific binding of gRNAs, three gRNAs were used and their on-target and commonly observed off-target sites (Tsai, S.Q. Ampli-sequencing of et al. Nat Biotechnology 33,187-197 (2015)., Kleinstiber, BP et al. Nature 529,490-495 (2016)., FIG. 11) was performed. Off-target risk scores were calculated for various base editing methods based on ampli-consequencing data (FIG. 12). The off-target risk score of the dual-function base editor was within the off-target risk score of the single-function base editor. In contrast, the off-target risk score for the base editor mixture was significantly higher than the risk score for the corresponding dual-function base editor. In addition, genome-wide DNA and RNA off-target activity of 13 base editors were extensively investigated by whole exome sequencing (WES) and transcriptome sequencing (RNA-seq). No elevated levels of single nucleotide polymorphism (SNV) were detected in the WES dataset, but the number of C → U RNA edits found using rAPOBEC1-related base editing methods rather than PmCDA1-related was previously reported ( Compared to the case of other samples (322) that match Grunewald, J. et al. Nature 569, 433-437 (2019)., Grunewald, J. et al. Nat Biotechnology 37, 1041-1048 (2019).). It was significantly higher (FIGS. 13A and 13B). Non-specific A → I RNA editing activity can be achieved by other previously reported methods (Grunewald, J. et al. Nature 569, 433-437) when using ABEmax or a base editing mixture containing ABE or ABEmax. (2019)., Zhou, C. et al. Nature 571,275-278 (2019)., Rees, HA, Wilson, C., Doman, J.L. & Liu, D.R.Sci Adv 5 , Eax5717 (2019).), Which was significantly higher overall (P value = 0.00019). In particular, the non-specific A → I editing activity of Target-ACEmax (mean 3,359 over two replications) was relatively lower than Target-AIDmax + ABEmax (mean 4,179).

Recent machine learning approaches to predict the outcome of genome editing via wild-type Cas9 (Shen, MW et al. Nature 563, 646-651 (2018)., Allen, F. et al. Nat Biotechnology, 64. -72 (2018)., Chen, W. et al. Nucleic Acids Res, gkz487 (2019).), The present inventors trained amplicon sequencing data and edited the base of a predetermined target sequence. We have developed a base editing prediction method for predicting patterns and their frequencies (Fig. 14). Simply put, our method predicts the base edit results for untrained targets with Pearson correlation coefficients of 0.70 and 0.71, respectively, for Target-ACEmax and ACBEmax. I succeeded in doing so. In addition, the ampli-consequencing data obtained in this example proved to be sufficient for the training procedure. Since the machine learning method can predict multinucleotide co-editing, it was used to predict all frequencies of codon conversion patterns in the human genome obtained by various base editing methods. If bystander mutations were not allowed to occur, this analysis showed that Target-ACEmax and its corresponding base editor mix, Target-AIDmax + ABEmax, have the highest potential for diversifying genomic codons. The same analysis was then repeated, allowing bystander mutations to occur. Then, the bystander risk of causing undesired mutations was estimated for all base editing methods (Fig. 15). As a result, the risk of bystander mutations in Target-ACEmax and ACBEmax was within that range of BE4max, which is the highest in the single-function base editor. The bystander mutation risk score for ACBEmax was significantly lower than that for Target-ACEmax, which was in close agreement with ACBEmax showing no significant improvement in the expansion of the genome-wide codon conversion pattern.

In addition, according to the model by the present inventors, Target-ACEmax obtained a pair of heterologous disease mutations reported in the ClinVar database (Landrum, MJ et al. Nucleic Acids Res 44, D862-868 (2016).). It was predicted that it was most likely to be corrected (Fig. 16).

As described above, according to the present invention, various base substitutions can be performed at the same time. Therefore, the present invention is expected to have a wide range of applications in the fields of agriculture, forestry and fisheries, livestock, medical care, etc., including breeding and gene therapy.

Claims (9)

1. A fusion protein in which different nucleobase converting enzymes are bound to the N-terminal and C-terminal of the Cas protein, and the Cas protein loses part or all of its nuclease activity.
2. The fusion protein according to claim 1, wherein the different nucleobase converting enzymes are adenosine deaminase and cytidine deaminase.
3. The fusion protein according to claim 2, wherein in the fusion protein, adenosine deaminase is bound to the N-terminal of the Cas protein and cytidine deaminase is bound to the C-terminal of the Cas protein.
4. The fusion protein according to any one of claims 1 to 3, further to which at least one of a nuclear localization signal and a uracil glycosylase inhibitor is bound.
5. A polynucleotide encoding the fusion protein according to any one of claims 1 to 4.
6. The polynucleotide according to claim 5, wherein the codon is optimized according to the host cell to be introduced.
7. An expression vector containing the polynucleotide according to claim 5 or 6.
8. A genome editing system including the following (A) and (B).
   (A) The fusion protein according to any one of claims 1 to 4, the polynucleotide according to claim 5 or 6, or the expression vector according to claim 7.
   (B) A guide RNA, a polynucleotide encoding the guide RNA, or an expression vector containing the polynucleotide.
9. A method for producing a cell in which DNA has been edited, which comprises introducing the genome editing system according to claim 8 into the cell.

Images