Retinal cell-targeting adeno-associated virus variant adn and use thereof

Abstract

Disclosed are adeno-associated virus (AAV) capsid protein variants with improved infectivity and transduction efficiency into target cells, and uses thereof. Specifically, disclosed are an AAV2 variant capsid protein selected through directed evolution for gene delivery to retinal cells, a recombinant AAV vector comprising same, and uses as a gene delivery system to retinal cells.

Description

Adeno-associated virus variant ADN targeting retinal cells and its uses

[0001] An adeno-associated virus (AAV) capsid protein variant with improved infectivity to target cells and transduction efficiency, and its use are disclosed. Specifically, an AAV2 variant capsid protein selected for gene delivery to retinal cells through design and validation using an AI-based model, a recombinant AAV vector containing the same, and its use as a gene delivery vehicle to retinal cells are disclosed.

[0002]

[0003] This research was conducted with support from the Samsung Future Technology Development Project (Project No. SRFC-MA2022-08).

[0004]

[0005] Gene therapy is a method of treating hereditary diseases by introducing normal genes into human cells to correct or compensate for diseases caused by genetic defects and abnormalities, and it is becoming a key means of treating various hereditary diseases.

[0006] Adeno-associated viruses (AAVs) are widely used as gene delivery vectors in the field of gene therapy because they can infect a wide range of tissues, are non-pathogenic, and have low immunogenicity. Various serotypes of AAV exist, and characteristics such as host or tissue specificity, transmissibility, and gene delivery efficiency differ depending on the serotype. For example, serotype 2 (AAV2) can infect various cells, and serotypes 1 (AAV1), serotype 5 (AAV5), and serotype 6 (AAV6) have higher tissue specificity compared to AAV2; thus, it has been reported that AAV1 has high gene delivery efficiency in muscle, liver, airway, and central nervous system; AAV5 in the central nervous system, liver, and retina; and AAV6 in the heart, muscle, and liver.

[0007] However, to be utilized as a gene delivery vector for gene therapy, it is necessary to possess higher tissue specificity and gene delivery and expression efficiency than existing AAV vectors. To this end, AAV capsid engineering, which involves modifying the sequence of the AAV capsid protein, is being actively researched.

[0008] Directed evolution is a representative high-throughput screening technique for improving the function of biomolecules, mimicking the process of natural selection by inducing genetic mutations and repeating selection processes. AAV capsid directed evolution proceeds by constructing a library of AAV capsid sequences through the introduction of mutations into wild-type AAV capsid genes, and then screening and securing capsid variants with desired characteristics from this library. However, there are limitations in that the generation and screening of numerous variant candidates require enormous amounts of time and cost.

[0009] Accordingly, research utilizing artificial intelligence (AI) has recently been actively conducted in the field of AAV capsid engineering, and methods are being utilized to secure capsid variants by selecting promising capsid variant candidates in silico through AI-based prediction models and verifying them through actual experiments.

[0010] Recombinant AAV vectors with enhanced tissue specificity and gene delivery efficiency have been developed through AAV capsid engineering and are being applied to various indications (e.g., Korean Registered Patent No. 2234930). In particular, hereditary retinal diseases are attracting attention as promising indications for gene therapy as they are often caused by single-gene abnormalities; to this end, variants such as AAV2-7m8, AAV2.GL, and AAV2.NN have been developed and are being utilized for gene delivery to the retina.

[0011] However, there is still a need for AAV variants with excellent transduction capabilities that can induce highly specific delivery to specific cells within the retina and induce gene expression.

[0012] Accordingly, the inventors constructed an artificial intelligence-based prediction model to secure AAV capsid variants that were not experimentally discovered by the AAV capsid screening method using a random library, and completed the present invention by using this model to design and verify AAV capsid variants with excellent gene delivery efficiency to target cells in the retina.

[0013] The purpose is to provide an AAV2 variant capsid protein capable of gene delivery to retinal cells via intravitreal injection.

[0014] In addition, the invention aims to provide a recombinant AAV2 vector containing an AAV2 variant capsid protein capable of gene transfer to retinal cells via intravitreal injection.

[0015] In addition, the invention aims to provide a therapeutic use for ocular diseases of a recombinant AAV2 vector containing an AAV2 variant capsid protein capable of gene delivery to retinal cells via intravitreal injection.

[0016]

[0017] One aspect of the present invention provides an AAV2 variant capsid protein comprising a peptide NAPPKNP inserted into the GH-loop of a wild-type AAV2 capsid protein.

[0018]

[0019] In order to discover AAV capsid sequences that were not experimentally identified in AAV screening using a random library, the inventors utilized artificial intelligence to screen and verify variants of the AAV2 capsid protein, thereby obtaining the AAV2 variant capsid protein of the present invention.

[0020] Specifically, data regarding the sequences of wild-type AAV2 capsid proteins selected from human retinal organoids and their gene delivery capabilities to photoreceptor cells were obtained. Based on this data, an artificial intelligence was used to train the correlation between the sequences and the gene delivery capabilities to photoreceptor cells. Based on this, a model was constructed to predict the sequences of AAV2 capsid variants predicted to exhibit superior gene delivery capabilities to photoreceptor cells. Using the prediction model constructed in this manner, candidate AAV2 capsid variant sequences were designed and applied to actual photoreceptor cells to verify gene delivery activity, thereby obtaining the AAV2 variant capsid protein of the present invention.

[0021] In one embodiment of the present invention, the peptide NAPPKNP may be inserted at amino acid 587 or a corresponding position of the wild-type AAV2 capsid protein of SEQ ID NO. 1.

[0022] The AAV2 variant capsid protein of the present invention includes the insertion of NAPPKNP between amino acids N587 and R588 of the wild-type AAV2 capsid protein of SEQ ID NO. 1, and was confirmed to deliver the target gene with high efficiency by transduction into retinal cells including photoreceptor cells, ganglion cells, Müller cells, and bipolar cells by intravitreal injection compared to AAV2.7m8 (which has the insertion of the peptide "LGETTRP" between N587 and R588 of wild-type AAV2: SEQ ID NO. 3) and AAV2.NN (which has the insertion of the peptide "NNPTPSR" between N587 and R588 of wild-type AAV2: SEQ ID NO. 4), which are known to have the ability to deliver genes to wild-type AAV2 and retinal photoreceptor cells, and was named "AAV2.ADN" or "ADN variant".

[0023]

[0024] The ADN variant according to the present invention was confirmed to pass through the inner limiting membrane (ILM) of the retina via intravitreal injection, deliver target genes to photoreceptor cells, and induce strong gene expression; it was also confirmed to pass through the retinal layer compared to wild-type AAV2, deliver to deep cells, and bring about increased expression of target genes; and it was confirmed to possess excellent gene delivery efficiency in human retinal organoids through increased gene expression and photoreceptor-specific gene delivery compared to wild-type AAV2.

[0025]

[0026] The wild-type AAV2 capsid protein contains the amino acid sequence of SEQ ID NO. 1. The ADN variant according to the present invention contains NAPPKNP between N587 and R588 in the amino acid sequence of SEQ ID NO. 1.

[0027] In one embodiment of the present invention, the AAV2 variant capsid protein may be composed of an amino acid sequence having at least 85%, at least 90%, or at least 95% sequence identity with respect to SEQ ID NO. 2.

[0028]

[0029] As used herein, the term "AAV" is an abbreviation for adeno-associated virus and refers to the virus itself or its derivatives. Unless otherwise specified, AAV includes all its subtypes and naturally occurring and recombinant forms. AAV is a non-pathogenic parvovirus consisting of a 4.7 kb-long single-stranded DNA genome within a non-encapsulated icosahedral capsid. The genome contains three open translation frames (ORFs) adjacent to an inverted terminal repeat (ITR) that functions as a viral origin for replication and packaging signals: the rep ORF codes for four non-structural proteins that play a role in viral replication, transcriptional regulation, site-specific integration, and virion assembly; and the cap ORF codes for three structural proteins (VP1-3) that assemble to form a 60-mer viral capsid; The ORF, which exists as an alternative reading frame within the cap gene, localizes the AAV capsid protein to the nucleolus and codes for the assembly-activating protein (AAP), a viral protein that functions during the capsid assembly process.

[0030] Genome sequences of various serotypes of AAV, sequences of natural terminal repeats (TRs), Rep proteins, and capsid subunits are known in the art. These sequences can be found in the literature or in public databases such as GenBank. For example, the genome sequence of AAV2 is registered as GenBank Accession NC_001401.

[0031] The term "tropism" as used herein refers to the preferential targeting of cells of a specific host species or specific cell types within a host species by a virus (e.g., AAV), and is used interchangeably with "tropism." For example, a virus capable of infecting retinal cells with high specificity has high tropism toward retinal cells compared to non-retinal cells. Since the capsid protein of an AAV virus determines infectivity toward target cells or tissues, tropism or tropism is used as a selection pressure to develop AAV variants capable of delivering genes to target cells or tissues with high specificity and efficiency through capsid engineering via directed evolution.

[0032] The term “retinal cell” as used herein means all types of cells present in the retina, including, but not limited to, photoreceptor cells, retinal ganglion cells, bipolar cells, Muller glia cells, horizontal cells, amacrine cells, astrocytes, microglia, retinal blood vessel cells, and retinal pigment epithelium (RPE).

[0033] As used herein, the term "gene delivery efficacy" refers to the ability or activity of an AAV vector to specifically deliver and express genes in target cells or tissues, and is used interchangeably with "infectiousness" or "transduction efficacy."

[0034] As used herein, the term “corresponding location” means a location that corresponds to a specific location through sequence alignment, or a location determined to perform the same or substantially the same function.

[0035] As used herein, the term "sequence identity" refers to the ratio of the type of amino acid or base at a specific position matching between polypeptides or polynucleotides being compared, and may be used interchangeably with "homology" for sequences. For example, the calculation of the homology (%) of two polypeptide sequences involves optimally aligning the two sequences for comparison purposes and comparing the amino acids at corresponding positions; if an amino acid identical to that at the corresponding position of the second sequence exists at the position of the first sequence, the two sequences are determined to be identical at that position. Sequence homology can be determined using software such as BLASTP, BLASTN, and FASTA.

[0036]

[0037] In one embodiment of the present invention, the AAV2 variant capsid protein can impart increased infectivity to retinal cells compared to the wild-type AAV2 capsid protein.

[0038] In one embodiment of the present invention, the retinal cells may be selected from the group consisting of photoreceptor cells, retinal ganglion cells, bipolar cells, Müller cells, horizontal cells, amacrine cells, astrocytes, microglia, retinal blood vessel cells, and retinal pigment epithelial cells.

[0039] In one embodiment of the present invention, the AAV2 variant capsid protein can impart increased infectivity to photoreceptor cells compared to the wild-type AAV2 capsid protein by intravitreal injection.

[0040]

[0041] Another aspect of the present invention provides an isolated nucleic acid encoding an AAV variant capsid protein.

[0042] In one embodiment of the present invention, the isolated nucleic acid may comprise a nucleotide sequence encoding the amino acid sequence of SEQ ID NO. 2 or a nucleotide sequence having 95% or more sequence homology with respect to it.

[0043]

[0044] Another aspect of the present invention provides a host cell comprising isolated nucleic acid encoding an AAV2 variant capsid protein comprising a peptide NAPPKNP inserted at amino acid 587 or a corresponding position of the wild-type AAV2 capsid protein of SEQ ID NO. 1.

[0045] The above host cell may be used to generate a recombinant AAV virion containing an AAV2 variant capsid protein. In this case, the host cell is referred to as a "packaging cell." The host cell may be selected from various cells including, but not limited to, HEK293 cells, HEK293T cells, HeLa cells, Sf9 cells, 293A cells, 293H cells, etc.

[0046] In one embodiment of the present invention, the isolated nucleic acid is stably or transiently introduced into a host cell using a method including, but not limited to, electroporation, calcium phosphate precipitation, liposome-mediated transfection, etc.

[0047]

[0048] Another aspect of the present invention is an AAV2 variant capsid protein comprising a peptide NAPPKNP inserted at amino acid 587 or a corresponding position of the wild-type AAV2 capsid protein of SEQ ID NO. 1, and

[0049] A recombinant AAV vector is provided, comprising a heterogeneous nucleic acid containing a sequence encoding a target product.

[0050]

[0051] As used herein, the term "recombinant AAV vector" refers to an AAV vector containing a polynucleotide sequence that is not derived from AAV, that is, a polynucleotide sequence heterogeneous to AAV, and is used interchangeably with "rAAV". Generally, in a recombinant AAV vector, the heterogeneous polynucleotide sequence is a nucleic acid sequence encoding the target product to be delivered to the target cell.

[0052] As used herein, the term "heterogeneity" refers to a source genetically distinct from the AAV2 vector, for example, nucleic acids originating from a different species.

[0053]

[0054] In one embodiment of the present invention, the recombinant AAV vector may exhibit increased transduction into retinal tissue / cells compared to an AAV vector (AAV2 vector) containing wild-type AAV2 capsid protein.

[0055] In one embodiment of the present invention, the recombinant AAV vector may exhibit increased transduction into retinal tissue / cells through intravitreal injection compared to an AAV vector (AAV2 vector) containing a wild-type AAV2 capsid protein.

[0056] A recombinant AAV vector according to one embodiment of the present invention can be used to produce a recombinant AAV virion. When introduced into a suitable host cell, the recombinant AAV vector can provide for the production of the corresponding recombinant AAV virion.

[0057] In one embodiment of the present invention, the target product may be a polypeptide or a nucleic acid. The target product refers to a product capable of producing a therapeutic effect on the target disease by supplementing or enhancing a function that is deficient or lacking in the target, or by reducing or eliminating a defective function. For example, the target product may be a gene product directly or indirectly related to an ocular disease such as a retinal disease, or one that alters the activity thereof, or one that improves the function of retinal cells, and may be, for example, PEDF (pigment epithelium-derived factor) or rhodopsin.

[0058] In one embodiment of the present invention, the target product may be a therapeutic nucleic acid and may target any gene associated with an ocular disease such as color blindness, age-related macular degeneration (such as wet AMD or dry AMD), cataract, panchoroidal atrophy, glaucoma, optic neuropathy, Marfan syndrome, myopia, nodular choroidal vasculopathy, retinitis pigmentosa, Stargardt disease, Usher syndrome, Leber congenital glaucoma, Leber hereditary optic neuropathy, uveal melanoma or X-linked retinal detachment, or any hereditary disease affecting ocular tissue. Target genes are, for example, VEGFA, VEGFB, ANGPT2, FLT1, KDR, PDGFB, CFH, ARMS2, HTRA1, TIMP3, C3, C2, CFB, SERPINF1, ANGPT1, CFI, CD59, ABCA1, ABCA4, RHO, RPGR, PRPF3, PRPF31, PRPF8, PDE6A, PDE6B, PDE6G, CNGB1, CNGA1, AIPL1, CRX, CRB1, CEP290, RS1, USH2A, MYO7A, USH1C, CLRN1, BBS1, BBS10, BEST1, PROM1, ELOVL4, PRPH2, RDH12, RPE65, CDH3, PCDH15, GUCY2D, GUCA1A, TULP1, LRAT, GNAT2, CNGB3, CNGA3, PDE6C, PDE6H, ATF6, GRM6, GPR179, TRPM1, NYX, CABP4, NR2E3, NRL, IMPDH1, IMPDH2, SNRNP200, HK1, GNAT1, KCNV2, RAX2, CHM, MERTK, EYS, NMNAT1, RP1, RP2, LCA5, RD3, RLBP1, MT-ND4, MT-ND1, MT-ND6, NDP, LRP5, TSPAN12, FZD4, CAPN5, TEK, CYP1B1, CYP4V2, MYOC, OPTN, OPA1, IL10, IL1RN, TNFRSF1A, HGF, PEDF, sFLT1, sTNFR, GNAQ, GNA11,BAP1, SF3B1, EIF1AX, TP53, TNFSF10, DDEF1, CDKN2A, p14ARF, IFNB1, HSV-TK, BDNF, GDNF, CNTF, NTF3, NGF, WDR87, APE1, MIP, GJA3, GJA8, CRYAA, CRYBB2, PRX, POLR3B, XRCC1, ZNF350, EPHA2, REP1, CALM2, MPP7, LOX1, CAV1, CAV2, PITX2, FOXC1, PAX6, LTBP2, FBN1, TGFBR2, MTHFR, MTR, MTRR, MET, UMODL1, It may be a gene encoding a protein selected from the group consisting of MMP1, MMP2, CBS, IGF1, UHRF1BP1L, PTPRR, PPFIA2, P4HA2, SERPING1, FGD6, ABCG1, LOC387715, CETP, PTEN, and HERC2 / OCA2, but is not limited thereto.

[0059] In one embodiment of the present invention, the target product may be a therapeutic protein or polypeptide, for example, VEGFA, VEGFB, ANGPT2, FLT1, KDR, PDGFB, CFH, ARMS2, HTRA1, TIMP3, C3, C2, CFB, SERPINF1, ANGPT1, CFI, CD59, ABCA1, ABCA4, RHO, RPGR, PRPF3, PRPF31, PRPF8, PDE6A, PDE6B, PDE6G, CNGB1, CNGA1, AIPL1, CRX, CRB1, CEP290, RS1, USH2A, MYO7A, USH1C, CLRN1, BBS1, BBS10, BEST1, PROM1, ELOVL4, PRPH2, RDH12, RPE65, CDH3, PCDH15, GUCY2D, GUCA1A, TULP1, LRAT, GNAT2, CNGB3, CNGA3, PDE6C, PDE6H, ATF6, GRM6, GPR179, TRPM1, NYX, CABP4, NR2E3, NRL, IMPDH1, IMPDH2, SNRNP200, HK1, GNAT1, KCNV2, RAX2, CHM, MERTK, EYS, NMNAT1, RP1, RP2, LCA5, RD3, RLBP1, MT-ND4, MT-ND1, MT-ND6, NDP, LRP5, TSPAN12, FZD4, CAPN5, TEK, CYP1B1, CYP4V2, MYOC, OPTN, OPA1, IL10, IL1RN, TNFRSF1A, HGF, PEDF, sFLT1, sTNFR, GNAQ, GNA11, BAP1, SF3B1, EIF1AX, TP53, TNFSF10, DDEF1, CDKN2A, p14ARF, IFNB1, HSV-TK, BDNF, GDNF, CNTF, NTF3, NGF, TAFIA, WDR87, APE1, MIP, GJA3, GJA8, CRYAA, CRYBB2, PRX, POLR3B, XRCC1, ZNF350, EPHA2,It may be selected from the group consisting of REP1, CALM2, MPP7, LOX1, CAV1, CAV2, PITX2, FOXC1, PAX6, LTBP2, FBN1, TGFBR2, MTHFR, MTR, MTRR, MET, UMODL1, MMP1, MMP2, CBS, IGF1, UHRF1BP1L, PTPRR, PPFIA2, P4HA2, SERPING1, FGD6, ABCG1, LOC387715, CETP, PTEN, and HERC2 / OCA2, but is not limited thereto.

[0060]

[0061] Another aspect of the present invention provides the use of the aforementioned recombinant AAV vector in the treatment of ocular diseases.

[0062] In one embodiment of the present invention, a pharmaceutical composition for treating an eye disease is provided, comprising the aforementioned recombinant AAV vector.

[0063] In one embodiment of the present invention, a method for treating an eye disease is provided, comprising the step of administering a therapeutically effective amount of the aforementioned recombinant AAV vector to a subject who requires treatment for an eye disease.

[0064] As used herein, the term "ocular disease" refers to a disease related to the eye or a part or specific area of ​​the eye, such as the retina or choroid.

[0065] In one embodiment of the present invention, the eye disease may be a retinal disease, a choroidal disease, or a retinal / choroidal disease.

[0066] In one embodiment of the present invention, the ocular disease may be selected from retinal vascular disease, optic neuropathy, hereditary retinal disease, hereditary choroidal disease, and intraocular tumor.

[0067] In one embodiment of the present invention, the ocular disease may be selected from retinal-vitreous-choroidal related disease, retinal vascular disease, macular disease, hereditary retinal disease, hereditary vitreous disease, hereditary choroidal disease, intraocular inflammatory disease, intraocular tumor, glaucoma, and optic neuropathy.

[0068] In one embodiment of the present invention, ocular diseases include age-related macular degeneration, diabetic retinopathy, retinal vascular occlusion such as retinal artery / vein occlusion, and macular telangiectasia; optic neuropathy including glaucoma; Hereditary retinal / choroidal diseases including uveitis, Leber congenital amaurosis, retinitis pigmentosa, Usher syndrome, Stargardt's Disease, Best's Disease, X-linked retinoschisis, congenital stationary night blindness, choroideremia, achromatopsia, cone dystrophy, gyrate atrophy, and Bardet-Biedl syndrome; intraocular tumors such as uveal melanoma, intraocular lymphoma, and retinoblastoma, but are not limited thereto.

[0069] In one embodiment of the present invention, the ocular disease may be selected from macular degeneration, diabetic retinopathy, retinal vascular occlusion, macular telangiectasia, retinopathy of prematurity, myopic degeneration, retinitis pigmentosa, Leber congenital amaurosis, Best's disease, Stargardt's disease, congenital non-progressive night blindness, X-linked retinal detachment, Vietti's crystal retinopathy, total color blindness, cone cell dystrophy, cone-rod cell dystrophy, macular dystrophy, Usher syndrome, Bardett-Beadle syndrome and other syndromic retinitis pigmentosa, panchoroidal atrophy, centrochoroidal dystrophy, cerebral choroidal atrophy, glaucoma, other optic neuropathy, uveitis, uveal melanoma, intraocular lymphoma, retinoblastoma, retinal detachment, and other retinal damage.

[0070]

[0071] In one embodiment of the present invention, the recombinant AAV vector may be administered via any suitable route of administration for delivery to the retina, such as intravitreal injection, subretinal injection, ocular injection by choroidal injection, or periocular, intravenous, arterial, or nasal administration, but not limited thereto.

[0072] In one embodiment of the present invention, the recombinant AAV vector may be administered by intravitreal injection. When the recombinant AAV vector of the present invention is administered via intravitreal injection, it can move through the vitreous humor, pass through the ILM, and move through the retinal layer more efficiently than an AAV vector containing wild-type AAV2 capsid protein.

[0073] In one embodiment of the present invention, the recombinant AAV vector may be administered by intraocular injection.

[0074] The recombinant AAV vector according to the present invention has an increased ability to deliver and express heterogeneous nucleic acids across the ILM to retinal cells compared to an AAV2 vector containing a wild-type capsid protein when administered via intravitreal injection.

[0075] The recombinant AAV vector according to the present invention can cross the ILM and also reach photoreceptor cells by traversing a cell layer including Müller cells, bipolar cells, etc. For example, when administered via intravitreal injection, the recombinant AAV vector according to the present invention can cross the ILM and also reach photoreceptor cells by traversing a cell layer including Müller cells, bipolar cells, etc.

[0076] In some embodiments, the recombinant AAV vector specifically infects retinal cells, and in particular, can specifically infect photoreceptor cells.

[0077]

[0078] As used herein, the term "therapeutic effective dose" refers to an amount sufficient to alleviate (e.g., relieve, reduce, or mitigate) at least one of the symptoms associated with a diseased state. For example, the therapeutic effective dose for in vivo injection, i.e., direct injection into the retina, is approximately 10 6 to about 10 15 , for example, about 10 8 to 10 12 It may be within the range of GC (genome copy). For in vitro transduction, the effective amount of rAAV virion delivered to cells is approximately 10 8 to about 10 13 The therapeutic effective dose of the recombinant AAV vector according to the present invention can be easily determined by a person skilled in the art by taking into account the target disease, age, gender, height, and severity of the disease to be treated.

[0079] In one embodiment of the present invention, a recombinant viral vector is administered in an amount sufficient to induce infection (or transduction) and expression of a heterogeneous nucleic acid sequence in the retinal cells of the subject.

[0080] In one embodiment of the present invention, the retinal cells may be selected from the group consisting of retinal ganglion cells, bipolar cells, photoreceptor cells, Müller glial cells, horizontal cells, amacrine cells, astrocytes, microglia, retinal blood vessel cells, and retinal pigment epithelial (RPE) cells.

[0081] In one embodiment of the present invention, the retinal cells are photoreceptor cells, e.g., cone cells and / or rod cells.

[0082] In one embodiment of the present invention, the retinal cell is a retinal ganglion cell (RGC).

[0083] In one embodiment of the present invention, the retinal cell is an RPE cell.

[0084] In one embodiment of the present invention, the pharmaceutical composition may be administered one or more times over periods of various intervals, for example, daily, weekly, monthly, or annually, to achieve a desired level of gene expression.

[0085] In one embodiment of the present invention, the pharmaceutical composition may include a pharmaceutically acceptable carrier, excipient, or additive.

[0086]

[0087] Another aspect of the present invention provides an in vitro method for delivering a heterogeneous nucleic acid encoding a target product to a retinal cell, comprising the step of contacting the aforementioned recombinant AAV vector with the retinal cell.

[0088] According to the present invention, delivering heterogeneous nucleic acids to retinal cells can be used for the treatment of retinal diseases.

[0089] In a specific embodiment of the present invention, the target product and retinal cells are as described above.

[0090]

[0091] The ADN variant according to the present invention has been confirmed to have excellent gene delivery efficiency by passing through the ILM via intravitreal injection, delivering target genes to retinal cells including photoreceptor cells, inducing strong gene expression, providing increased gene expression and lateral spreading compared to wild-type AAV2, and exhibiting increased photoreceptor-specific gene delivery and expression compared to wild-type AAV2 in human retinal organoids. Therefore, the ADN variant according to the present invention can be usefully employed in gene therapy for the treatment of ocular diseases.

[0092]

[0093] Figure 1 shows a schematic diagram of an AI-based AAV capsid engineering method that includes the process of learning AAV variant selection data using a deep learning-based model and designing AAV variants based thereon. Specifically, AAV capsid data capable of delivering to retinal cells from human retinal organoids is collected (Data collection), and data on AAV characteristics, such as the AAV2 capsid sequence and gene delivery ability to photoreceptor cells, is obtained through Next Generation Sequencing (NGS). Subsequently, the obtained data is preprocessed to be suitable for training a deep learning model (Data preprocessing), and the correlation between the sequence and AAV characteristics is learned through AI (Training). Next, the relationship between amino acids at each position of the sequence and AAV characteristics is analyzed through the trained model (Analyze), and based on this, a model capable of predicting AAV2 characteristics for the AAV2 capsid sequence is constructed (Predictiction), and an AAV2 capsid sequence with optimal characteristics can be designed using the constructed model (Design). By transfecting designed AAV2 capsid sequences into human retinal organoids and verifying their gene delivery ability to photoreceptor cells, we selected AAV2 variant capsid proteins that demonstrated improved transfection into retinal cells compared to AAVs with wild-type AAV2 capsid.

[0094] Figure 2 shows the performance evaluation of deep learning models (2D CNN, 1D CNN, and DNN) for photoreceptor directivity for selecting AAV2 variants according to one embodiment of the present invention. Three models based on 2D CNN, 1D CNN, and DNN algorithms were constructed by training AAV2 capsid sequence data selected from photoreceptor cells after preprocessing by one-hot encoding, and all three models showed decent learning performance, and it was confirmed that the prediction performance was also decent in the parity plot between actual values ​​and predicted values.

[0095] Figure 3 shows a vp3 structure image of AAV2.ADN containing an ADN variant according to one embodiment of the present invention predicted by ColabFold. The part indicated by the dotted circle represents the inserted heptameric peptide.

[0096] Figures 4a and 4b show the gene delivery ability of AAV2.ADN containing an ADN variant according to one embodiment of the present invention verified in human retinal organoids. After transgenerating recombinant AV2.ADN (SEQ No. 2), wild-type AAV2 (SEQ No. 1), improved AAV2.7m8 (SEQ No. 3), and AAV2.NN (SEQ No. 4) loaded with the GFP (green fluorescent protein) gene as the target gene into human retinal organoids (CRX-tdTomato), it was confirmed that AAV2.ADN exhibited superior gene expression efficiency throughout the retinal organoid compared to wild-type AAV2 and improved AAV2.7m8 and AAV2.NN. Figure 4a is a low-magnification image, and Figure 4b is a high-magnification image.

[0097] Figures 5a and 5b show the results of a quantitative analysis of the gene transfer efficiency of AAV2.ADN according to one embodiment of the present invention confirmed in human retinal organoids. tdTomato fluorescence labels photoreceptor cells, and GFP fluorescence labels the cells to which the gene was transferred. In Figure 5b, the gene transfer efficiency for all retinal cells (Total) was analyzed as the ratio of GFP-positive cells to the total cells, the gene transfer efficiency to photoreceptor cells was analyzed as the ratio of GFP-positive cells to tdTomato-positive cells, and the gene transfer efficiency to non-photoreceptor retinal cells was analyzed as the ratio of GFP-positive cells to tdTomato-negative cells. AAV2.ADN showed significantly higher gene delivery efficiency than wild-type AAV2 and AAV2.7m8 in retinal organoid photoreceptor cells, and showed higher gene delivery efficiency than wild-type AAV2, improved AAV2.7m8, and AAV2.NN in whole cells and non-photoreceptor cells, respectively, and this was statistically significant.

[0098] Figure 6 shows the results of immunofluorescence staining indicating the gene transfer efficiency of recombinant AAV2 vectors loaded with a gene encoding GFP as a target gene in AAV2.ADN, wild-type AAV2, improved AAV2.7m8, and AAV2.NN according to one embodiment of the present invention. Antibodies labeling photoreceptor cells and Müller cells (CRX and Recoverin antibodies: photoreceptor cells, CARLBP and SOX9: Müller cells) and antibodies against GFP fluorescence were used.

[0099] Figure 7 shows the results of verifying the in vivo gene delivery efficiency of AAV2.ADN according to one embodiment of the present invention. It shows the expression patterns of the target gene observed after intravitreal administration of AAV2.ADN, loaded with the tdTomato gene as the target gene, and wild-type AAV2, improved AAV2.7m8, and AAV2.NN as controls to reporter mice (Nrl-EGFP) labeled with EGFP fluorescence in rod cells. In the figure, RGC refers to retinal ganglion cells, INL refers to the inner nuclear layer, ONL refers to the outer nuclear layer, and OS refers to the outer segment. While wild-type AAV2 was confirmed to infect RGC and some INL, it was confirmed that the tdTomato gene delivered by AAV2.ADN was expressed at a significantly high level not only in RGC and INL but also in the ONL and OS regions, which are the nuclear layers of photoreceptor cells.

[0100] Figure 8 shows the results of a quantitative analysis of the in vivo gene transfer efficiency of AAV2.ADN according to one embodiment of the present invention. It shows the expression patterns of the target gene confirmed by FACS analysis after intravitreal administration of AAV2.ADN loaded with the tdTomato gene as the target gene, and wild-type AAV2, improved AAV2.7m8, and AAV2.NN as controls, respectively, into mice (Nrl-EGFP). EGFP fluorescence labels rod cells, and tdTomato fluorescence labels the cells to which the gene was transferred. The gene transfer efficiency to all retinal cells (Total) was measured as the ratio of tdTomato-positive cells to the total cells, the gene transfer efficiency to rod cells was measured as the ratio of tdTomato-positive cells to EGFP-positive cells, and the gene transfer efficiency to retinal cells other than rod cells was measured as the ratio of tdTomato-positive cells to EGFP-negative cells. AAV2.ADN showed significantly higher gene transfer efficiency in rod cells compared to wild-type AAV2 and improved types AAV2.7m8 and AAV2.NN, and this was statistically significant.

[0101]

[0102] The present disclosure will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present disclosure is not limited to the embodiments and may be implemented in various forms. The embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined by the claims.

[0103]

[0104] Example 1. Prediction and Design of AAV Variants Using Artificial Intelligence

[0105] Artificial intelligence-based AAV2 capsid engineering was performed to design and select AAV2 capsid variants capable of efficiently delivering and expressing target genes in retinal cells. Specifically, to discover AAV capsid sequences that were not experimentally identified during AAV selection using a random library, a deep learning model trained on data selected from human retinal organoids was analyzed using an XAI (explainable artificial intelligence) algorithm, and AAV variants capable of effectively delivering genes to human retinal cells were designed. A schematic diagram of artificial intelligence-based AAV capsid engineering is shown in Figure 1.

[0106] Data on AAV capsid delivery capabilities to retinal cells were collected from human retinal organoids. Data on AAV2 capsid sequences and AAV characteristics, such as gene delivery capabilities to retinal photoreceptors, were obtained via NGS. Subsequently, artificial intelligence was used to train the correlation between AAV capsid sequences and AAV characteristics, such as directivity toward retinal photoreceptors. Through this process, a model for designing AAV capsid sequences capable of predicting AAV capsid sequences with desired characteristics was constructed. This model analyzes the influence of amino acids at specific positions within the AAV capsid sequence on AAV characteristics and, based on this analysis, predicts the AAV capsid sequence with optimal characteristics. Specifically, by visualizing the amino acids and their positions that influence AAV characteristics, the model analyzes the combination of amino acid composition and resulting AAV characteristics to provide predicted values ​​for the performance of the AAV capsid sequence, thereby enabling the selection of the optimal AAV capsid sequence.

[0107]

[0108] 1.1. Data Collection by Random Library Screening Using Human Retinal Organoids

[0109] (1) Production of human retinal organoids in vitro

[0110] Human retinal organoids to be used to evaluate the gene delivery efficacy of AAV2 variants into retinal cells were prepared according to the description in Wahlin, KJ et al. Photoreceptor Outer Segment-like Structures in Long-Term 3D Retinas from Human Pluripotent Stem Cells. Sci Rep 7,766 (2017). Specifically, human retinal organoids were prepared using human embryonic stem cells that specifically express tdTomato fluorescent protein in photoreceptor cells. Human embryonic stem cells that specifically express tdTomato fluorescent protein in retinal cells were cultured in mTesR1 basal medium (Stem Cell Technologies, Vancouver, Canada) on Matrigel-coated plates (Corning, NY, USA). Before inducing differentiation into retinal organoids, the cultured stem cells were cultured for 4-5 days. After dissociating the cultured stem cells using Accutase (BD Biosciences, CA, USA), they were aggregated in low-cell adhesion 96-well plates (Corning) with U-shaped bottoms at a density of 3,000 cells per well. Subsequently, human retinal organoids were obtained by differentiating and culturing for more than 30 weeks. All cultures for the preparation of human retinal organoids were maintained at 37°C in a humidified environment with 5% carbon dioxide.

[0111]

[0112] (2) Construction of a random library

[0113] A random library was constructed by inserting random heptameric peptides into the GH loop (containing amino acids 587 and 588 of the AAV2 capsid of SEQ ID NO. 1), which is known as the cellular interaction domain of wild-type AAV2. In this case, the theoretical diversity that the AAV random library can possess is 1.28 × 10⁶. 9 However, the diversity of AAV random libraries that can actually be implemented is 1×10 6-7 To overcome these limitations of the AAV random library, we aimed to design an optimal human retinal cell-targeted AAV variant by utilizing artificial intelligence to learn the relationship between the AAV capsid sequence and the human retinal cell target.

[0114]

[0115] (3) Selection of variants capable of transferring genes to photoreceptor cells

[0116] - Screening

[0117] 5x10 random libraries of heptmeric peptides inserted into human retinal organoids labeled with tdTomato fluorescence in fabricated photoreceptor cells 10 Variants with high gene transfer efficiency to retinal cells were selected by transduction via GC (Genome Copy) and FACS separation using tdTomato fluorescence after 3 weeks.

[0118]

[0119] - Cell separation and NGS via FACS

[0120] For Fluorescence-activated cell sorting (FACS), human retinal organoids were dissociated into single cells using the Papain Dissociation System (Worthington Biochemical Corporation, NJ, USA). FACS was performed using the BD FACSymphony S6 Cell Sorter (BD Biosciences, CA, USA). For human retinal organoids, tdTomato-positive and tdTomato-negative cells were separated.

[0121] After recovering the isolated cells by centrifugation, DNA was extracted using the QIAamp DNA Mini Kit (QIAGEN, 51306). PCR was performed to amplify the 229 bp of the extracted DNA containing the inserted heptameric peptide, and the target DNA was obtained as a PCR product using the QIAquick Gel Extraction Kit (QIAGEN, 51306). Next Generation Sequencing (NGS) analysis of the PCR products was performed using the MiSeq Reagent Kits v2 (ILLUMINA, MS-102-2003) according to the manufacturer's instructions to collect AAV capsid data with gene delivery capability to retinal cells. The NGS data included the relationship between the infection rate of human retinal cells according to the AAV capsid sequence.

[0122]

[0123] 1.2. NGS Data Preprocessing for AI Model Training: Mean-Square Error (MSE)

[0124] To design an AAV2 variant capsid capable of delivering genes to retinal cells, sequence information of AAV variant capsids infected with human retinal cells obtained via NGS in 1.1 was used. Since the AAV capsid sequence information is string data and artificial intelligence can only accept numeric data as input, a process of converting the string-based AAV capsid sequence data into numeric data was required.

[0125] One-hot encoding was utilized for conversion into numerical data, and one-hot encoding requires setting a labeling order. Training performance was evaluated using two types of labeling orders: one based on alphabetical order and the other based on the chemical properties of amino acids. Low overfitting and MSE were observed in the one-hot encoding based on chemical properties, leading to the establishment of a labeling order based on these characteristics (basic, acidic, polar, and non-polar).

[0126]

[0127] The data obtained from NGS represents the number of copies of the AAV variant capsid sequences that infected each cell. AAV variants that account for a large proportion of the randomly generated library used may result in a large absolute number of infected copies, even if their infection efficiency in human retinal cells is low. To correct for this, the commonly used log enrichment score (output copy number ratio / input copy number ratio) was introduced and established as the score. The proportion of individual variants in the AAV library recovered from photoreceptor cells was adjusted to the proportion of individual variants in the injected AAV library. This allows for the relative comparison of human photoreceptor targeting capabilities among AAV variants based on changes in the proportion of individual variants within the AAV library.

[0128]

[0129] 1.3. Construction of a Deep Learning Model for Photoreceptor Directivity

[0130] A deep learning model was trained using the enrichment score as the output and the recovered AAV capsid sequence as the input. Since it is necessary to learn the abstract characteristics of AAV contained within the AAV capsid sequence, training was conducted using deep learning techniques specialized in learning non-linearity and abstract information. To enhance the reliability of the training, three deep learning models were constructed based on 2D CNN, 1D CNN, and DNN algorithms. Based on the data preprocessing method described in Section 1.2, data selected from photoreceptor cells was preprocessed, and the models were trained. Specifically, the correlation between the AAV2 capsid sequence that reached the photoreceptor cells, the number of copies of the sequence, and the directivity toward the photoreceptor cells was trained. All three models demonstrated decent training performance, and it was confirmed that their prediction performance was also satisfactory through the partiy plot (actual value, predicted value). The results are illustrated in Figure 2.

[0131]

[0132] The trained deep learning model includes weights for 20 amino acids at different positions of the heptmeric peptide regarding the relationship between the AAV capsid sequence and the infection rate of human retinal cells. Using XAI techniques, the weights learned by the deep learning model are visualized and analyzed to suggest amino acids capable of effectively infecting human retinal cells. Based on this, amino acid combinations that have a positive effect on human retinal cell infection in common across the three models were selected, and 'NAPPKNP' (hereinafter AAV2.ADN), which is predicted to have excellent gene delivery ability to photoreceptor cells, was selected. This variant was named the "ADN variant."

[0133]

[0134] The amino acid sequence of the capsid protein of AAV2.ADN, in which the ADN variant (indicated in red) is inserted at amino acid 587 of the AAV2 wild-type capsid protein (SEQ No. 1) (SEQ No. 2), is as follows:

[0135]

[0136] In addition, the vp3 structure image of AAV2.ADN containing an ADN variant was confirmed using ColabFold (Nature Methods, 19(6), 679-682). Figure 3 shows the vp3 structure image of AAV2.ADN containing an ADN variant according to one embodiment of the present invention, predicted by ColabFold.

[0137]

[0138] Example 2. In vitro verification of AAV2.ADN variants: Human retinal organoids

[0139] To verify the ability of the AAV2.ADN variant selected in Example 1 to be delivered and expressed into photoreceptor cells, the delivery and expression of the target gene were confirmed in human retinal organoids.

[0140]

[0141] 2.1. Confirmation of Gene Transfer and Expression by Fluorescence Microscopy

[0142] Recombinant vectors AAV2.ADN-EGFP, wild-type AAV2-EGFP, positive controls AAV2.NN-EGFP, and AAV2.7m8-EGFP (see Science Translational Medicine, 5(189), 189ra76 and EMBO Molecular Medicine, 13(4), e13392) loaded with the gene encoding EGFP were applied to photoreceptor reporter (CRX-tdTomato) organoids that specifically express tdTomato fluorescence as described in Example 1, 5x10 10 Gene expression efficiency was confirmed after transduction with GC.

[0143]

[0144] In human retinal organoids, photoreceptor cells, which are the target cells of AAV2.ADN, are located at the outermost layer. When a gene expressing EGFP was delivered, it was confirmed that, as is known, wild-type AAV2 could not penetrate the retina layer and thus could not infect the interior. Furthermore, in human retinal organoids treated with AAV2.ADN, the region appearing yellow due to the merging of photoreceptor-specific tdTomato fluorescent protein and EGFP fluorescent protein was wider than in human retinal organoids treated with AAV2, AAV2.7m8, and AAV2.NN. Therefore, AAV2.ADN demonstrated significantly higher EGFP gene delivery efficiency throughout the human retinal organoid compared to wild-type AAV2 and the improved versions AAV2.7m8 and AAV2.NN. The results are illustrated in Figures 4a and 4b.

[0145] The AAV2.ADN variant was confirmed to have superior gene delivery to human photoreceptor cells compared to wild-type AAV2, AAV2.7m8, and AAV2.NN.

[0146]

[0147] 2.2. Quantitative Analysis of Gene Transfer Efficiency by FACS

[0148] The gene expression efficiency and transduction efficiency of AAV2.ADN in human retinal organoids were quantitatively analyzed.

[0149]

[0150] Specifically, AAV2.ADN loaded with a gene encoding EGFP as the target gene and control wild-type AAV2, improved AAV2.7m8, and AAV2.NN were loaded onto human retinal organoids expressing tdTomato fluorescence specifically in photoreceptor cells, prepared as described in 1.1., 5x10 10Human retinal organoids were transfected with vg, and FACS analysis was performed after 3 weeks. Human retinal organoids were dissociated using the Papain Dissociation System (Worthington Biochemical Corporation, NJ, USA). FACS was performed using BD FACS Aria II (BD Biosciences, CA, USA), and analysis was performed using FlowJo software (Tree Star, OR, USA).

[0151]

[0152] tdTomato fluorescence labels photoreceptor cells, and GFP fluorescence labels cells to which the target gene has been delivered. The gene delivery efficiency to all retinal cells was analyzed as the ratio of GFP-positive cells to total cells, the gene delivery efficiency to photoreceptor cells was analyzed as the ratio of GFP-positive cells to tdTomato-positive cells, and the gene delivery efficiency to retinal cells other than photoreceptor cells was analyzed as the ratio of GFP-positive cells to tdTomato-negative cells.

[0153] According to FACS analysis, AAV2.ADN had a higher proportion of GFP-positive cells than wild-type AAV2, AAV2.7m8, and AAV2.NN used as controls, and among them, the proportion of GFP-positive cells was higher among photoreceptor cells (tdTomato positive). The proportion of GFP-positive cells among all human retinal organoid cells in AAV2.ADN was 80.88 ± 3.70%, which was significantly higher and statistically significant than that of wild-type AAV2 (10.19 ± 1.09%), improved AAV2.7m8 (39.08 ± 4.26%), and AAV2.NN (63.76 ± 3.05%) used as controls. In addition, the proportion of GFP-positive cells among photoreceptor cells (tdTomato+) was 66.86 ± 1.68% for the selected AAV2.ADN, which was significantly higher and statistically significant than that of AAV2 (14.31 ± 1.84%) and AAV2.7m8 (46.01 ± 2.37%). The results are illustrated in Figures 5a and 5b, respectively, as a representative result image of the FACS analysis and a graph showing the results of repeated FACS analysis.

[0154]

[0155] 2.3. Detection of Gene Expression by Immunofluorescence Staining

[0156] To confirm gene transfer to photoreceptor cells and Muller glia cells, immunofluorescence staining was performed using antibodies labeling these cells and antibodies against EGFP delivered by AAV2.

[0157]

[0158] Human retinal organoids transfected with AAV2.ADN-EGFP and the control wild-type AAV2-EGFP, improved AAV2.NN-EGFP, and AAV2.7m8-EGFP, respectively, were fixed in a 5% sucrose solution containing 4% paraformaldehyde for 25 minutes. After fixation, they were sequentially treated in 10%, 20%, and 30% sucrose solutions for 1 hour each. For sectioning, human retinal organoids were placed using an OCT (Optical Coherence Tomography) compound and rapidly frozen using liquid nitrogen vapor. Subsequently, 7 μm thick sections were prepared using a cryostat (Leica Microsystems, Wetzlar, Germany) to confirm differences in GFP gene expression in the human retinal organoids. The results are shown in Figure 7. The expression of the GFP gene delivered to AAV2 variants and wild-type AAV2 merged with the fluorescence of photoreceptor cell labeling antibodies (CRX and Recoverin antibodies) and Müller glial cell labeling antibodies (CRALBP and SOX9 antibodies), respectively. The GFP gene expression induced by the AAV2.ADN variant had a wider merged region in photoreceptor cells and Müller glial cells than that of wild-type AAV2 and improved AAV2.NN and AAV2.7m8. Therefore, it was confirmed that the selected AAV2.ADN had a higher gene delivery efficiency to photoreceptor cells and Müller glial cells than that of wild-type AAV2 and improved AAV2.7m8 and AAV2.NN.

[0159]

[0160] Example 3. Phosphoric in vivo verification of AAV2.ADN variant: Mouse

[0161] The gene delivery ability of the AAV2.ADN variant selected in Example 1 to photoreceptor cells was confirmed in reporter mice (B6.Cg-Tg(Nrl-EGFP)1Asw / J) with EGFP fluorescence labeled on rod photoreceptors.

[0162]

[0163] 3.1. Confirmation of Gene Transfer and Expression by Fluorescence Microscopy

[0164] To verify the gene delivery efficiency of AAV2.ADN in mice, 8-week-old mice (B6.Cg-Tg(Nrl-EGFP)1Asw / J) expressing EGFP fluorescence in rod cells were treated with AAV2.ADN loaded with the gene encoding tdTomato and AAV2 / AAV2.7m8 / AAV2.NN as a control, each at a rate of 1 x 10⁶. 10 GC was injected into the vitreous cavity, and after 3 weeks, mice were sacrificed and their eyes were enucleated. The enucleated eyes were fixed in 1% paraformaldehyde for 1 hour, and then sequentially treated in 10%, 20%, and 30% sucrose solutions for 1 hour each. For section preparation, the eyes were positioned using an OCT (Optical Coherence Tomography) compound to ensure horizontal alignment of the optic nerve and the central corneal junction, rapidly frozen using liquid nitrogen vapor, and 7 µm thick sections were prepared using a cryostat (Leica Microsystems, Wetzlar, Germany). After treatment with DAPI, a cell nucleus label, the expression of EGFP and tdTomato was imaged using a fluorescence microscope.

[0165]

[0166] It was confirmed that wild-type AAV2 infected retinal ganglion cells (RGCs) and some inner nuclear layers (INL). In contrast, it was confirmed that AAV2.ADN significantly increased the expression of the tdTomato gene, which was delivered not only to RGCs and INLs but also to the outer nuclear layer (ONL) and outer segment (OS) regions of photoreceptor cells. The results are shown in Figure 6.

[0167]

[0168] 3.2. Quantitative Analysis of Gene Transfer Efficiency by FACS

[0169] To quantify the gene delivery efficiency of AAV2.ADN in mice, 8-week-old mice (B6.Cg-Tg(Nrl-EGFP)1Asw / J) with EGFP fluorescence-labeled rod cells were treated with AAV2.ADN loaded with the gene encoding tdTomato and controls of AAV2, AAV2.7m8, and AAV2.NN, each at a rate of 1 x 10⁶ 10 Intravitreal injection was performed via GC, and after 3 weeks, mice were sacrificed and eyes were enucleated. The cells were dissociated into single cells using the Papain Dissociation System (Worthington Biochemical Corporation, NJ, USA), and FACS was performed using the BD FACSymphony S6 Cell Sorter (BD Biosciences, CA, USA). FACS analysis was performed using FlowJo software (Tree Star, OR, USA).

[0170]

[0171] EGFP fluorescence labels rod cells, and tdTomato fluorescence labels cells to which the gene has been delivered. The gene delivery efficiency to total retinal cells was measured as the ratio of tdTomato-positive cells to total cells, the gene delivery efficiency to rod cells was measured as the ratio of tdTomato-positive cells to EGFP-positive cells, and the gene delivery efficiency to retinal cells excluding rods was measured as the ratio of tdTomato-positive cells to EGFP-negative cells. FACS analysis results showed that AAV2.ADN exhibited significantly higher gene delivery efficiency in rod cells than wild-type AAV2 and improved types AAV2.7m8 and AAV2.NN. The proportion of tdTomato-positive cells among rod cells (EGFP+) for AAV2.ADN was 38.50 ± 4.80%, which is significantly higher and statistically significant than that of AAV2 (6.52 ± 2.00%), AAV2.7m8 (15.95 ± 3.42%), and AAV2.NN (15.78 ± 3.09%). For AAV2.ADN, the gene transfer efficiency was higher than that of AAV2 in total retinal cells and retinal cells excluding rod cells, and this was statistically significant. The results are shown in Figure 8.

[0172]

[0173] Through these results, it was confirmed that the ADN variant selected in Example 1, compared to wild-type AAV2 and improved AAV2.7m8 and AAV2.NN, delivers the target gene to photoreceptor cells by passing through the ILM from the injection site at a high rate through intravitreal injection and expresses the target gene at a high rate.

Claims

1. An AAV2 variant capsid protein comprising the peptide NAPPKNP inserted into the GH-loop of a wild-type AAV2 capsid protein.

2. The AAV2 variant capsid protein of Claim 1, wherein the peptide is inserted at amino acid 587 or a corresponding position of the wild-type AAV2 capsid protein of SEQ ID NO.

1.

3. The AAV2 variant capsid protein of claim 1, wherein the AAV2 variant capsid protein confers increased infectivity to retinal cells through intravitreal injection compared to the wild-type AAV2 capsid protein.

4. The AAV2 variant capsid protein of claim 3, wherein the retinal cells are selected from the group consisting of photoreceptor cells, retinal ganglion cells, bipolar cells, Müller cells, horizontal cells, amacrine cells, astrocytes, microglia, retinal blood vessel cells, and retinal pigment epithelial cells.

5. Isolated nucleic acid encoding the AAV2 variant capsid protein of Claim 1.

6. A host cell comprising isolated nucleic acid encoding the AAV2 variant capsid protein of Claim 1.

7. AAV2 variant capsid protein of Claim 1, and A recombinant AAV vector comprising a heterogeneous nucleic acid containing a sequence encoding a target product.

8. The recombinant AAV vector of claim 7, wherein the recombinant AAV vector results in increased transduction into retinal cells via intravitreal injection compared to an AAV vector containing wild-type AAV2 capsid protein.

9. The recombinant AAV vector of claim 8, wherein the retinal cells are selected from the group consisting of photoreceptor cells, retinal ganglion cells, bipolar cells, Müller cells, horizontal cells, amacrine cells, astrocytes, microglia, retinal vascular cells, and retinal pigment epithelial cells.

10. A recombinant AAV vector according to claim 7, wherein the object product is a polypeptide or nucleic acid.

11. A pharmaceutical composition for treating ocular diseases comprising the recombinant AAV vector of claim 7.

12. A pharmaceutical composition according to claim 11, wherein the ocular disease is selected from retinal-vitreous-choroidal related disease, retinal vascular disease, macular disease, hereditary retinal disease, hereditary vitreous disease, hereditary choroidal disease, intraocular inflammatory disease, intraocular tumor, glaucoma, and optic neuropathy.

13. A pharmaceutical composition according to claim 11, wherein the eye disease is selected from macular degeneration, diabetic retinopathy, retinal vascular occlusion, macular telangiectasia, retinopathy of prematurity, myopic degeneration, retinitis pigmentosa, Leber congenital amaurosis, Best's disease, Stargardt's disease, congenital non-progressive night blindness, X-linked retinal detachment, Vietti's crystal retinopathy, total color blindness, cone cell dystrophy, cone-rod cell dystrophy, macular dystrophy, Usher syndrome, Bardett-Beadle syndrome, syndromic retinitis pigmentosa, panchoroidal atrophy, centrochoroidal dystrophy, cerebral choroidal atrophy, glaucoma, optic neuropathy, uveitis, uveal melanoma, intraocular lymphoma, retinoblastoma, retinal detachment, and retinal injury.

14. The pharmaceutical composition of claim 11, wherein the pharmaceutical composition is administered by intravitreal injection, subretinal injection, or macula injection.

15. An in vitro method for delivering heterogeneous nucleic acids to retinal cells, comprising the step of contacting the recombinant AAV vector of claim 7 with retinal cells.

16. The method of claim 15, wherein the retinal cells are selected from the group consisting of photoreceptor cells, retinal ganglion cells, bipolar cells, Müller cells, horizontal cells, amacrine cells, astrocytes, microglia, retinal blood vessel cells, and retinal pigment epithelial cells.

Images