Preventive or therapeutic agent for neurodegenerative disease

Abstract

To provide a method for preventing or treating a neurodegenerative disease by predicting the most crucial neurodegenerative death during the period in which neurological symptoms manifest and the disease progression occurs in the neurodegenerative disease, identifying a molecular target for treating the neurodegenerative disease, and inhibiting the expression of the identified molecular target.SOLUTION: A preventive or therapeutic agent for neurodegenerative disease comprising an agent for suppressing KLHL32 gene expression, and a screening method for a preventive or therapeutic agent for neurodegenerative disease, the method comprising (1) a step of bringing a test substance into contact with cells that express the KLHL32 gene, and (2) a step of selecting, as a candidate for the preventive or therapeutic agent for neurodegenerative disease, a test substance that has reduced the expression level of the KLHL32 gene.SELECTED DRAWING: None

Description

Technical Field

[0001] The present invention relates to a preventive or therapeutic agent for neurodegenerative diseases, which contains an agent for suppressing the expression of the KLHL32 gene.

Background Art

[0002] Alzheimer's disease (AD) is a degenerative disease of the brain that represents dementia, and the number of its patients is constantly increasing. The decline in the quality of life (QoL) associated with the decline in brain nerve function in Alzheimer's disease patients has a great impact on not only the patients themselves but also their families and the like, and has become a serious problem facing modern society. In June 2021, the US Food and Drug Administration (FDA) conditionally approved "aducanumab", a candidate for a therapeutic agent for Alzheimer's disease. On the other hand, the FDA has requested Biogen, one of the developers, to conduct a follow-up investigation to confirm whether aducanumab is useful for treating the symptoms of Alzheimer's disease.

[0003] Similar to Alzheimer's disease, frontotemporal lobar degeneration (FTLD) is known as a disease showing progressive neurodegenerative disorders. Frontotemporal lobar degeneration is the second or third most frequent early-onset neurodegenerative dementia after Alzheimer's disease, shows symptoms of prominent behavioral and personality changes, and often accompanied by language dysfunction, which gradually develops into cognitive impairment and dementia. Also, similar to Alzheimer's disease, research is being carried out, but the whole picture of the pathogenesis has not yet been clarified.

[0004] Parkinson's disease is a representative neurodegenerative disease comparable to Alzheimer's disease, and it is known that Lewy bodies appear in the substantia nigra of patients' brains. Lewy bodies are aggregates of a protein consisting of 140 amino acid residues called α-synuclein, and are also known to appear in the brains of patients with diseases called Lewy body diseases such as Lewy body dementia and multiple system atrophy in addition to Parkinson's disease. Diseases accompanied by the accumulation of α-synuclein in the brain are called α-synucleinopathies. It is thought that the aggregation of α-synuclein, that is, α-synuclein fibril formation, plays a significant role in the progression of α-synucleinopathy, and research is actively underway in various fields on substances that suppress α-synuclein fibril formation and the like.

[0005] However, although research on these neurodegenerative diseases is being vigorously advanced, the full picture of their pathogenesis is not clear, and at present, a radical cure for them has not been developed. As a common pathological condition of many neurodegenerative diseases including Alzheimer's disease, neurodegeneration accompanied by the activation of caspase 3 can be cited (for example, Non-Patent Document 1). The present inventors previously developed a method for predicting the presence and degree of neurodegeneration from phase-contrast images using deep learning and induced pluripotent stem cell (iPS cell) technology (Patent Document 1). However, the mechanism of this prediction method and what the deep learning model used as an indicator for predicting neurodegeneration still remain unclear.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Non-Patent Documents

[0007]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0008] Therefore, an object of the present invention is to predict the most important neurodegeneration at the time when neurological symptoms in neurodegenerative diseases become apparent and the disease condition progresses, and to identify molecular targets for treating neurodegenerative diseases. And by suppressing the expression of the identified molecular target, to provide a method for preventing or treating neurodegenerative diseases.

Means for Solving the Problems

[0009] The present inventors constructed a model subjected to deep learning, and by this model, for cells with a high score (cells with a high probability of being cells in which caspase-3 is expected to be activated) and cells with a low score (cells with a low probability of being cells in which caspase-3 is expected to be activated), single cell RNA sequencing (single cell RNA-seq) analysis was performed. As a result, KLHL32 was found as a gene specifically expressed before the activation of caspase-3 in the cell group with a high score. Next, when analyzing the autopsy brain of Alzheimer's disease patients, the expression of KLHL32 was specifically recognized at the site where β-amyloid accumulates, and it was confirmed that in the Alzheimer's disease mouse model, it is expressed in neurons prior to the activation of caspase-3. Furthermore, in the cell model of Alzheimer's disease, it was revealed that when the expression of the KLHL32 gene was knocked down by siRNA, the activation of caspase-3 was suppressed and cell death was suppressed. As a result of further research based on these findings, the present inventors completed the present invention.

[0010] That is, the present invention is as follows. [1] A prophylactic or therapeutic agent for neurodegenerative diseases, comprising an agent for suppressing the expression of the KLHL32 gene. [2] The prophylactic or therapeutic agent according to claim 1, wherein the expression inhibitor is selected from the group consisting of siRNA, heteroduplex nucleic acid, antisense nucleic acid, shRNA, miRNA, antigene nucleic acid, and CRISPR-Cas system. [3-1] The prophylactic or therapeutic agent according to [2], wherein the expression inhibitor is siRNA. [3-2] The prophylactic or therapeutic agent according to [2], wherein the CRISPR-Cas system is a CRISPR-dCas system. [4] The prophylactic or therapeutic agent according to any one of [1] to [3-2], wherein the neurodegenerative disease is selected from the group consisting of Alzheimer's disease, Parkinson's disease, amyotrophic lateral sclerosis (ALS), spinocerebellar degeneration, frontotemporal lobar degeneration, Lewy body dementia, multiple system atrophy, Huntington's disease, progressive supranuclear palsy, and corticobasal degeneration. [5] The prophylactic or therapeutic agent according to any one of [1] to [4], wherein the neurodegenerative disease is Alzheimer's disease. [6] (1) A step of contacting a test substance with a cell that expresses the KLHL32 gene, and (2) A step of selecting a test substance with a reduced expression level of the KLHL32 gene as a candidate for a prophylactic or therapeutic agent for neurodegenerative diseases A screening method for a prophylactic or therapeutic agent for neurodegenerative diseases, comprising: [7] (1) A step of contacting a test substance with a cell, and (2) A step of selecting a test substance that suppresses an increase in the expression level of the KLHL32 gene as a candidate for a prophylactic or therapeutic agent for neurodegenerative diseases A screening method for a prophylactic or therapeutic agent for neurodegenerative diseases, comprising: [8] The method according to [6] or [7], wherein the cell is a model cell for neurodegenerative diseases. [9] The method according to [8], wherein the model cells of the neurodegenerative disease are cells derived from a patient with a neurodegenerative disease or cells differentiated from pluripotent stem cells having a mutation in a gene that causes the neurodegenerative disease.

[10] The method according to any one of [6] to [9], wherein the cells are nerve cells.

[11] The method according to any one of [6] to

[10] , wherein the cells are human cells.

[12] A biomarker for diagnosing a neurodegenerative disease, comprising a KLHL32 protein or a KLHL32 transcript.

[13] A method for assisting in the diagnosis of whether a subject has a neurodegenerative disease, comprising the step of detecting the biomarker according to

[12] in a sample from the subject or the subject.

[14] A kit for diagnosing a neurodegenerative disease, comprising an antibody that specifically recognizes a KLHL32 protein or a nucleic acid probe or nucleic acid primer that specifically recognizes a KLHL32 transcript.

[15] A method for preventing or treating a neurodegenerative disease in a mammal, characterized by administering an effective amount of an agent for suppressing the expression of the KLHL32 gene to the mammal.

[16] An agent for suppressing the expression of the KLHL32 gene for use in preventing or treating a neurodegenerative disease.

[17] Use of an agent for suppressing the expression of the KLHL32 gene for the manufacture of a prophylactic or therapeutic agent for a neurodegenerative disease.

[18] A method for preventing or treating a neurodegenerative disease, comprising the following steps (i) to (iii). (i) A step of detecting the biomarker according to

[12] in a sample from a subject or the subject; (ii) A step of diagnosing whether the subject has a neurodegenerative disease based on the result of step (i); (iii) A step of administering a prophylactic or therapeutic agent for a neurodegenerative disease to a subject diagnosed as having a neurodegenerative disease in step (ii).

[19] The method according to

[18] , wherein the preventive or therapeutic agent for the neurodegenerative disease is selected from the group consisting of an agent for suppressing the expression of the KLHL32 gene, aducanumab, donepezil, rivastigmine, galantamine, donepezil, and memantine.

Effects of the Invention

[0011] According to the present invention, it becomes possible to predict neurodegeneration significantly prior to neurodegeneration, and further, a preventive or therapeutic agent for a neurodegenerative disease based on a new molecular target (KLHL32 gene), a method for assisting diagnosis, and a method for screening the agent are provided.

Brief Description of the Drawings

[0012]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Modes for Carrying Out the Invention

[0013] 1. A preventive or therapeutic agent for neurodegenerative diseases As shown in the following examples, in a cell model of Alzheimer's disease (AD), the inventors have demonstrated that when the expression of the KLHL32 gene is knocked down by siRNA, the activation of caspase-3 is suppressed and cell death is inhibited. Neurodegeneration associated with the activation of caspase-3 is considered to be a common pathological condition of all neurodegenerative diseases including AD. Therefore, a drug that suppresses the expression of the KLHL32 gene can be used as a preventive or therapeutic drug for neurodegenerative diseases in general including AD. That is, the present invention provides a preventive or therapeutic drug for neurodegenerative diseases (hereinafter sometimes referred to as "the drug of the present invention") comprising a drug that suppresses the expression of the KLHL32 gene. In addition, unless otherwise specified, a preventive or therapeutic drug (or method) for a neurodegenerative disease also includes a drug (or method) that can prevent and treat the disease.

[0014] The drug of the present invention can be administered orally or parenterally as a pharmaceutical composition or a preventive or therapeutic agent in an appropriate dosage form by using the drug that suppresses the expression of the KLHL32 gene, which is an active ingredient, alone as it is or by mixing it with a pharmacologically acceptable carrier, excipient, diluent, etc. Further, the drug of the present invention can be administered to mammals (e.g., humans, rats, mice, guinea pigs, rabbits, sheep, horses, pigs, cows, dogs, cats, monkeys). Therefore, there is also provided a method for preventing or treating a neurodegenerative disease in a mammal, which comprises administering an effective amount of a drug that suppresses the expression of the KLHL32 gene to the mammal.

[0015] Since neurodegeneration associated with the activation of caspase-3 is considered to be a common pathological condition of all neurodegenerative diseases, in the present invention, the neurodegenerative diseases to be targeted for prevention or treatment are not particularly limited, and examples thereof include Alzheimer's disease (AD), Parkinson's disease, amyotrophic lateral sclerosis (ALS), spinocerebellar degeneration, frontotemporal lobar degeneration, Lewy body dementia, multiple system atrophy, Huntington's disease, progressive supranuclear palsy, corticobasal degeneration, etc. Among them, AD is preferred.

[0016] In the present invention, the AD to be targeted for prevention or treatment may be either sporadic or familial AD. In the case of familial AD, the causative gene is not particularly limited and can be any known causative gene including, for example, Amyloid Precursor Protein (APP), Presenilin 1 (PSEN1), Presenilin 2 (PSEN2), and the like. In one embodiment, in the case of familial AD having an APP mutation, the APP gene mutations include, but are not limited to, dup APP mutation, APP KM670 / 671NL mutation, APP D678N mutation, APP E682K mutation, APP A692G mutation, APP E693K mutation, APP E693Q mutation, APP E693G mutation, APP E693del (APP E693Δ) mutation, APP D694N mutation, APP L705V mutation, APP A713T mutation, APP T714A mutation, APP T714I mutation, APP V715M mutation, APP V715A mutation, APP I716V mutation, APP I716F mutation, APP I716T mutation, APP V717I mutation, and the like. Here, for example, APP E693Δ means a deletion-type mutation of E693 in APP.

[0017] In the present invention, the ALS to be targeted for prevention or treatment includes both sporadic and familial ALS. In the case of familial ALS, the causative gene is not particularly limited, and examples include SOD1, TDP-43, C9orf72, alsin, SETX, FUS / TLS, VAPB, ANG, FIG4, OPTN, ATXN2, DAO, UBQLN2, PFN1, DCTN1, CHPM2B, VCP, etc. In one embodiment, in the case of familial ALS having an SOD1 mutation, examples of the SOD1 gene mutation include a mutation in which Leu at position 144 of the SOD1 protein is substituted with Phe-Val-Xaa (Xaa represents any amino acid) (SOD1-L144FVX), a mutation in which Gly at position 93 is substituted with Ser (SOD1-G93S), a mutation in which Leu at position 106 is substituted with Val (SOD1-L106V), etc., but are not limited thereto. Further, the present invention can also be suitably used for patients having a mutation in the MAPT gene having a mutation in exons 9-13. Examples of the mutation in exons 9-13 include, for example, K257T, I260V, G272V, N297K, K280Δ, L284L, N296N, P301L, P301S, S305N, S305S, V337M, E342V, G389R, R406W, etc.

[0018] In this specification, the "therapeutic agent" includes not only pharmaceuticals aimed at the radical treatment of neurodegenerative diseases, but also, for example, pharmaceuticals aimed at suppressing the progression of these diseases, pharmaceuticals aimed at alleviating symptoms (for example, improving to minimal manifestations (MM) with no interference in life or work), or pharmaceuticals aimed at reducing sequelae. For example, since neurodegenerative diseases are diseases that progress over a long period (usually on a yearly basis), the progression of symptoms can be prevented by starting treatment early. Also, in this specification, the "preventive agent" includes not only pharmaceuticals aimed at reducing the risk of developing neurodegenerative diseases in subjects who have not developed neurodegenerative diseases, but also pharmaceuticals aimed at reducing the risk of recurrence of neurodegenerative diseases in subjects who have developed neurodegenerative diseases. For example, potentially, for patients with a genetic background that is susceptible to developing neurodegenerative diseases, the onset of neurodegenerative diseases can also be prevented by administering the pharmaceutical of the present invention before the symptoms of neurodegenerative diseases appear. The same applies to the "treatment method" and the "prevention method".

[0019] As a drug for suppressing the expression of the KLHL32 gene, it is not limited as long as it can suppress the expression of the gene at least in nerve cells, and the substance may be a nucleic acid. For example, antisense nucleic acids (e.g., antisense oligonucleotides (ASO), etc.) (including nucleic acids encoding such nucleic acids), siRNA (including nucleic acids encoding such siRNA), heteroduplex oligonucleotide (HDO), shRNA (including nucleic acids encoding such shRNA), miRNA (microRNA) (including nucleic acids encoding such miRNA), antigene nucleic acids, CRISPR-Cas systems, etc. may be mentioned. Hereinafter, when the drug for suppressing the expression of the KLHL32 gene is a nucleic acid, such a nucleic acid may be referred to as "the nucleic acid of the present invention". When the drug for suppressing the expression of the KLHL32 gene is a CRISPR-Cas system, the nucleic acids encoding the system (i.e., the nucleic acid encoding Cas, the nucleic acid encoding the guide RNA, and the nucleic acid encoding the guide RNA and Cas) and the guide RNA are also included in "the nucleic acid of the present invention". The medicament of the present invention may contain only one or two or more drugs for suppressing the expression of the KLHL32 gene. When two or more are contained, the same kind may be included (e.g., a plurality of siRNAs having different target sequences are included, etc.), or different kinds may be included (e.g., siRNA and antisense nucleic acid are included, etc.).

[0020] The KLHL gene generally belongs to the KLHL (Kelch-like) gene family that encodes a protein having a BTB / POZ domain, a BACK domain, and 5 to 6 Kelch motifs. As the human KLHL32 protein, a 620-amino acid-long protein containing a BTB / POZ domain, a BACK domain, and 5 Kelch motifs is exemplified. Molecular types of KLHL1 to KLHL42 are known in the KLHL family, and some of them are predicted to be involved in tumorigenesis and the E3-mediated proteolytic system in the endoplasmic reticulum, but the relationship between the KLHL32 protein and neurodegeneration has not been reported.

[0021] There are multiple isoforms of the KLHL32 protein. When an agent for suppressing the expression of the KLHL32 gene targets KLHL32 transcripts such as antisense nucleic acids, siRNA, heteroduplex nucleic acids, shRNA, or miRNA, if the subject of prevention or treatment is a human, the expression-suppressing agent only needs to be able to suppress the expression of the KLHL32 transcript encoding at least the full-length KLHL32 protein. Therefore, the expression-suppressing agent contains a base sequence complementary to a partial sequence of the KLHL32 transcript encoding the full-length KLHL32 protein (hereinafter referred to as the target RNA sequence). Such a base sequence can be designed, for example, based on the cDNA base sequence (SEQ ID NO: 1 or 2) of the KLHL32 transcript encoding the full-length human KLHL32 protein, which is registered as Genbank Accession No. NM_001323252.2 or NM_052904.4. When an agent for suppressing the expression of the KLHL32 gene targets the KLHL32 gene such as an antigene nucleic acid or the CRISPR / Cas system, the expression-suppressing agent (in the case of the CRISPR / Cas system, the guide RNA constituting the system) contains a base sequence complementary to a partial sequence of the KLHL32 gene (hereinafter referred to as the target DNA sequence). Such a base sequence can be designed, for example, based on the DNA base sequence of the human KLHL32 gene registered as NCBI Gene's Gene ID: 114792, location: NC_000006.12 (96898083..97145030). Specific examples of the target RNA sequence include, but are not limited to, the base sequences represented by any of SEQ ID NOs: 4 to 7.

[0022] The length of the target RNA sequence or target DNA sequence is not particularly limited as long as the KLHL32 gene expression inhibitor can specifically recognize and bind to the sequence, but is preferably 12 nucleotides or longer, more preferably 15 nucleotides or longer, still more preferably 17 nucleotides or longer. The upper limit of the length is also not particularly limited, but is, for example, 30 nucleotides or shorter, preferably 25 nucleotides or shorter, more preferably 22 nucleotides or shorter. Therefore, examples of the range of the length of the target region include 12 to 30 nucleotides, preferably 15 to 25 nucleotides, more preferably 17 to 22 nucleotides.

[0023] In the present specification, "expression of the KLHL32 gene" is used in the meaning including at least "production of KLHL32 transcript" unless otherwise specified, but is preferably used in the meaning further including "production of functional KLHL32 protein". Therefore, suppression of gene expression means not only that the abundance of the transcript transcribed from the gene in the cell decreases by contact with the KLHL32 gene expression inhibitor, but may also include that the abundance of the functional protein encoded by the gene in the cell decreases. Further, in the present specification, KLHL32 transcripts typically include KLHL32 mRNA and KLHL pre-mRNA, but preferably KLHL32 mRNA.

[0024] As used herein, "nucleic acid" may mean a monomeric nucleotide, but usually means a nucleotide consisting of a plurality of monomers (e.g., oligonucleotide, polynucleotide, etc.). Therefore, unless otherwise specified, when intending a monomeric nucleotide, it shall be denoted as "nucleic acid nucleotide", and examples of such nucleic acids include ribonucleic acid, deoxyribonucleic acid, peptide nucleic acid (PNA), morpholino nucleic acid, etc. Further, unless otherwise specified, when a nucleic acid is a nucleotide consisting of a plurality of monomers, each nucleotide residue constituting the nucleic acid (including the nucleotides at the 5'-end and 3'-end) shall simply be referred to as "nucleotide". Furthermore, as used herein, "nucleic acid strand" or "strand" shall mean a single-stranded nucleic acid unless otherwise specified. Therefore, "antisense strand" can be read as "single-stranded antisense nucleic acid".

[0025] As used herein, "complementary" means a relationship in which nucleobases can form so-called Watson-Crick base pairs (natural base pairs) or non-Watson-Crick base pairs (Hoogsteen-type base pairs, wobble base pairs, etc.) via hydrogen bonds. Therefore, "complementary sequence" is used in the sense of including not only a sequence that is completely complementary to a target RNA sequence or target DNA sequence (i.e., hybridizes without mismatch), but also a sequence containing 1 to several (e.g., 2, 3, 4, 5 or more) mismatches as long as it can hybridize to the target sequence under stringent conditions or under physiological conditions of mammalian cells. For example, sequences having 80% or more (e.g., 85% or more, 90% or more, 95% or more, 96% or more, 97% or more, 98% or more, 99% or more), most preferably 100% identity to a sequence that is completely complementary to a target RNA sequence or target DNA sequence are included.

[0026] Stringent conditions may be either low-stringent conditions or high-stringent conditions. Low-stringent conditions may be conditions of relatively low temperature and high salt concentration, for example, 30 °C, 2×SSC, 0.1% SDS. High-stringent conditions may be conditions of relatively high temperature and low salt concentration, for example, 65 °C, 0.1×SSC, 0.1% SDS. By changing conditions such as temperature and salt concentration, the stringency of hybridization can be adjusted. Here, 1×SSC contains 150 mM sodium chloride and 15 mM sodium citrate.

[0027] An antisense nucleic acid is a single-stranded nucleic acid containing a sequence complementary to a target RNA sequence. The antisense nucleic acid forms a double-stranded region with the target RNA sequence by the sequence complementary to the target RNA sequence, and suppresses the expression of the KLHL32 gene by the double-stranded region being cleaved by ribonuclease H (RNase H).

[0028] From the viewpoints of stability in vivo and efficient cleavage of transcripts, the antisense nucleic acid is preferably a gapmer-type nucleic acid. In the present invention, the "gapmer-type nucleic acid" means a chimeric antisense nucleic acid in which an internal region having a plurality (e.g., 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 or more) of nucleotides recognized by RNase H (hereinafter sometimes referred to as the "gap region" in this specification) is disposed between external regions having one or more (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 or more) nucleotides chemically modified so as to impart resistance to ribonuclease (hereinafter sometimes referred to as "modified nucleotides", and specific examples will be described later). At least one nucleoside constituting each of the 3'-wing region and the 5'-wing region is preferably a cross-linked nucleoside (specific examples will be described later).

[0029] The length of the antisense nucleic acid is not particularly limited, but is typically 10 to 50 nucleotides in length, preferably 10 to 30 nucleotides in length, more preferably 13 to 30 nucleotides in length, and even more preferably 15 to 20 nucleotides in length.

[0030] The antisense nucleic acid can be prepared by determining the sequence based on the sequence of the KLHL32 transcript and synthesizing a sequence complementary thereto using a commercially available DNA / RNA automatic synthesizer (such as Applied Biosystems, Beckman, etc.).

[0031] siRNA is a double-stranded RNA composed of an RNA having a sequence complementary to the target RNA sequence (i.e., the antisense strand) and its complementary strand. Further, a single-stranded RNA in which a sequence complementary to the target RNA sequence (the first sequence) and its complementary sequence (the second sequence) are linked via a hairpin loop portion, and by taking a hairpin loop type structure, the first sequence forms a double-stranded structure with the second sequence (small hairpin RNA: shRNA) is also one of the preferred embodiments of siRNA. Further, shRNA may be in the form of a nucleic acid (e.g., expression vector, etc.) encoding the shRNA.

[0032] HDO means a double-stranded nucleic acid composed of a main-chain DNA (i.e., an antisense strand) and an RNA (cRNA) strand complementary to the DNA. When taken into cells, the cRNA strand of HDO is cleaved by RNase H in the cells. The main-chain DNA is typically a gapmer-type nucleic acid, and the wing regions of the cRNA strand are typically composed of modified nucleotides such as 2'-O-methyl RNA (specific examples will be described later). Also, typically, all phosphodiester bond portions of the main-chain DNA and / or cRNA are sulfurized. It is presumed that the main-chain DNA, which has become a single-stranded antisense nucleic acid by cleavage of the cRNA strand, binds to the target RNA, and the antisense effect is exerted by RNase H cleaving the target RNA again. As long as the cRNA strand serves as a substrate for RNase H, the main-chain DNA and the cRNA strand may be modified other than those described above.

[0033] The double-stranded nucleic acid may have an overhang (also referred to as a protruding portion) at the 5'-end and / or 3'-end of one or both of the antisense strand or the complementary strand. The overhang is typically formed by the addition of 1 to several (e.g., 1, 2, 3, 4, 5, or 6), preferably 1 to 3 bases at the end of the sense strand and / or the antisense strand. Examples of the overhang include those consisting of dTdT or UU. The overhang can be present only in the antisense strand, only in the sense strand, and in both the antisense strand and the sense strand. However, in the present invention, a double-stranded nucleic acid having protruding portions in both the antisense strand and the sense strand is preferably used.

[0034] The length of each nucleic acid strand of the double-stranded nucleic acid is not particularly limited as long as the antisense effect can be exerted. For example, it is 10 to 50 nucleotides in length, preferably 15 to 30 nucleotides in length, and more preferably 20 to 27 nucleotides in length.

[0035] Double-stranded nucleic acids can be obtained by chemically synthesizing them using conventionally known methods or by producing them using genetic recombination techniques. It is also possible to use commercially available nucleic acids as appropriate. For example, double-stranded nucleic acids can be appropriately designed based on the target RNA sequence using commercially available software (e.g., RNAiDesigner; Invitrogen). The sense strand and the antisense strand of the target sequence on the RNA are each synthesized using a commercially available DNA / RNA automatic synthesizer (Applied Biosystems, Beckman, etc.), denatured at about 90°C to about 95°C for about 1 minute in an appropriate annealing buffer, and then annealed at about 30°C to about 70°C for about 1 to about 8 hours to prepare them. In addition, HDO can also be prepared by the method described in WO2013 / 089283, for example.

[0036] As used herein, "miRNA" refers to a single-stranded RNA or double-stranded RNA (e.g., miRNA / miRNA * ) that does not cleave the target RNA like siRNA but recognizes the 3' untranslated region (UTR) of the target RNA and controls translation. miRNA refers to an endogenous non-coding RNA (ncRNA) of about 20 to 25 bases originally encoded on the genome. Pri-miRNA is expressed from the miRNA gene, and then mature-miRNA is generated through the production of pre-miRNA. Thereafter, mature-miRNA is incorporated into RISC to generate single-stranded miRNA. The miRNA used in the present invention may be in the form of pri-miRNA, may be in the form of pre-miRNA, may be in the form of mature-miRNA (miRNA / miRNA * ), and may further be in the form of single-stranded RNA. In addition, the miRNA used in the present invention includes those that act in the nucleus and degrade the transcription product in an RNase H-dependent manner by a gapmer structure in which RNA oligomers are arranged at both ends and a DNA oligomer is arranged in the central part. In addition, miRNA may be in the form of a nucleic acid (e.g., expression vector, etc.) encoding the miRNA.

[0037] miRNA may be synthesized in the same manner as the method for synthesizing the aforementioned double-stranded nucleic acid. Further, miRNA can also be prepared by substituting the sequence of the miRNA according to the target RNA sequence based on any endogenous miRNA. Examples of such endogenous miRNAs include, but are not limited to, let-7, miR-15a, miR-143, miR-139, and their precursors.

[0038] The length of miRNA (when miRNA is double-stranded, the length of each nucleic acid strand) is not particularly limited as long as the antisense effect can be exerted. For example, it is 10 to 50 nucleotides in length, preferably 15 to 30 nucleotides in length, and more preferably 20 to 27 nucleotides in length.

[0039] An antisense nucleic acid refers to a nucleic acid that can suppress target gene expression at the transcriptional level by forming a triple strand with double-stranded DNA having the target gene. A typical antisense method is based on the binding of a third nucleic acid to a DNA double strand. The bases of the third nucleic acid form hydrogen bonds with the purine bases of the purine-pyrimidine base pairs, enabling triple strand formation by taking a structure in which three bases are arranged continuously in a plane. There are two types of the carbon position at the 1'-position of the base of the third nucleic acid (i.e., antisense nucleic acid): one with an outward orientation (Hoogsteen binding type) and the other with the opposite orientation (reverse Hoogsteen binding type). The former has a parallel orientation (the same orientation as the strand on the side where the base pairs of the DNA double strand are formed), and the latter has an antiparallel orientation (the opposite orientation to the strand on the side where the base pairs of the DNA double strand are formed). Parallel orientation TFO binds to T (T:A-T base pair) and protonated C (C+:G-C base pair) to A and G of the A-T base pair and G-C base pair in the DNA double strand, respectively. On the other hand, antiparallel orientation TFO binds to T (T:A-T base pair) and G (G:G-C base pair) to A and G of the A-T base pair and G-C base pair in the DNA double strand, respectively. Since C needs to be protonated under acidic conditions, its use under physiological conditions is limited. However, this problem can be avoided by using methylcytidine that can be protonated even under neutral conditions or derivatives that do not require protonation (e.g., 5-methyl-6-oxocytidine).

[0040] The nucleotide molecule constituting the antisense nucleic acid may be natural DNA or RNA, or may be a nucleotide subjected to various chemical modifications in order to improve stability (chemical and / or against enzymes) and specific activity (affinity with DNA).

[0041] The length of the antisense nucleic acid is not particularly limited, but is typically 10 to 50 nucleotides in length, preferably 10 to 30 nucleotides in length, more preferably 13 to 30 nucleotides in length, and even more preferably 15 to 20 nucleotides in length.

[0042] Antisense nucleic acids can be chemically synthesized, for example, by methods known per se or methods analogous thereto described in WO 2005 / 021570, WO 03 / 068695, WO 2001 / 007455.

[0043] The CRISPR / Cas system recognizes the target DNA sequence by a complex of a short-chain CRISPR RNA (crRNA) that forms a double strand with the target DNA sequence and a trans-activating crRNA (tracrRNA) (i.e., double-stranded guide RNA), or a single synthetic RNA (i.e., single-stranded guide RNA) combining crRNA and tracrRNA. Therefore, any sequence can be targeted by designing the sequence of crRNA based on the target DNA sequence. The CRISPR / Cas system is provided as a complex of single-stranded or double-stranded guide RNA and Cas (also referred to as Cas nuclease).

[0044] The design of the base sequence complementary to the target DNA sequence contained in the guide RNA can be appropriately designed, for example, by using a publicly available guide RNA design website (such as CRISPR Design Tool, CRISPRdirect, etc.). From among these candidates, a candidate sequence with a small number of off-target sites in the target host genome can be used as the targeting sequence. When the guide RNA design software used does not have a function to search for off-target sites in the host genome, for example, by performing a Blast search against the host genome for 8 to 12 nucleotides on the 3' side of the candidate sequence (a seed sequence with high discriminative ability for the target nucleotide sequence), off-target sites can be searched. tracrRNA, etc. can also be appropriately designed according to the type of Cas nuclease.

[0045] The Cas used in the present invention is not particularly limited, but is preferably Cas9 (also referred to as Cas9 nuclease) or Cpf1 (also referred to as Cpf1 nuclease). Examples of Cas9 include Cas9 derived from Streptococcus pyogenes (SpCas9; PAM sequence (in the 5'→3' direction; the same applies hereinafter) NGG (N is A, G, T, or C; the same applies hereinafter)), Cas9 derived from Staphylococcus aureus (SaCas9; PAM sequence NNGRR(T)), Cas9 derived from Streptococcus thermophilus (StCas9; PAM sequence NNAGAAW), Cas9 derived from Neisseria meningitidis (MmCas9; PAM sequence NNNNGATT), etc., but are not limited thereto. Preferably, it is SpCas9. Examples of Cpf1 include Cpf1 derived from Francisella novicida (FnCpf1; PAM sequence TTN), Cpf1 derived from Acidaminococcus sp. (AsCpf1; PAM sequence TTTN), Cpf1 derived from Lachnospiraceae bacterium (LbCpf1; PAM sequence TTTN), etc., but are not limited thereto.

[0046] Considering its use in human clinical applications, it is not preferable to cause DNA double-strand breaks (DSBs). Therefore, as Cas, those with inactivated DNA cleavage activity (dCas) are preferable. Thus, as the CRISPR-Cas system used in the present invention, a CRISPR-dCas system is preferable. As dCas, for example, in the case of SpCas9, a D10A mutant lacking the cleavage ability of the strand complementary to the strand forming a complementary strand with the guide RNA, in which the 10th Asp residue is converted to an Ala residue, and a double mutant of an H840A mutant lacking the cleavage ability of the complementary strand of the guide RNA, in which the 840th His residue is converted to an Ala residue, can be used. Also, in the case of FnCpf1, mutants lacking the cleavage ability of both strands, in which the 917th Asp residue is converted to an Ala residue (D917A) or the 1006th Glu residue is converted to an Ala residue (E1006A), can be used. Other mutant Cas can be used in the same manner.

[0047] When using the CRISPR / Cas system, a complex of guide RNA and Cas binds to the target DNA sequence, thereby inhibiting the transcription of the target gene. Also, by binding a transcriptional repressor to Cas, it is possible to further enhance the transcriptional repression of the target gene. Examples of such transcriptional repressors include KRAB, MBD2B, v-ErbA, SID (including a concatemer of SID (SID4X)), MBD2, MBD3, the DNMT family (e.g., DNMT1, DNMT3A, DNMT3B, etc.), Rb, MeCP2, ROM2, and AtHD2A, etc., and preferably, KRAB. When using KRAB as the transcriptional repressor of the present invention, the protein from which it is derived is not particularly limited, and examples include KOX-1 (ZNF10), KOX8 (ZNF708), ZNF43, ZNF184, ZNF91, HPF4, HTF10, HTF34, etc.

[0048] The CRISPR / Cas system can be included in the medicament of the present invention in the form of, for example, a complex of guide RNA and Cas, a complex of a nucleic acid encoding Cas and guide RNA, a complex of a nucleic acid encoding guide RNA and Cas, or a nucleic acid encoding guide RNA and Cas (which may be a single nucleic acid or multiple nucleic acids). These nucleic acids may be RNA such as mRNA, but are preferably DNA such as expression vectors.

[0049] As the expression vector used in the present invention, for example, viral vectors such as inactivated retrovirus, adenovirus, adeno-associated virus, herpes virus, vaccinia virus, poxvirus, poliovirus, sindbis virus, Sendai virus, SV40, immunodeficiency virus (HIV), etc. can be used. Preferably, adenovirus or adeno-associated virus vectors are mentioned.

[0050] The expression vector typically has a promoter upstream of the DNA encoding the nucleic acid or protein of the present invention. Such promoters include SRα promoter, SV40 promoter, LTR promoter, CMV (cytomegalovirus) promoter, RSV (Rous sarcoma virus) promoter, MoMuLV (Moloney murine leukemia virus) LTR, HSV-TK (herpes simplex virus thymidine kinase) promoter, etc. Also, a promoter specific to a particular cell (for example, a nerve cell) may be used.

[0051] The length of the base sequence complementary to the target DNA sequence contained in the guide RNA is not particularly limited as long as it can specifically bind to the target DNA sequence, but is, for example, 15 to 30 nucleotides, preferably 18 to 25 nucleotides.

[0052] The CRISPR / Cas system, the nucleic acid encoding the guide RNA, and the nucleic acid encoding Cas can also construct the DNA encoding the full-length protein by chemically synthesizing a DNA strand or a protein, or by connecting oligonucleotide short strands that are partially overlapping and synthesized using the PCR method or the Gibson Assembly method.

[0053] The nucleotide components constituting the nucleic acid of the present invention include a sugar moiety (e.g., ribose, deoxyribose), a base, and a phosphate. Also, examples of the "modification" include substitutions, additions, and / or deletions in the components and / or internucleoside linkages, and substitutions, additions, and / or deletions of atoms and / or functional groups in the components and / or internucleoside linkages.

[0054] Examples of natural bases include adenine, cytosine, guanine, thymine, and uracil. Also, examples of modified bases obtained by modifying the bases include, but are not limited to, 5-methylcytosine, 5-fluorocytosine, 5-bromocytosine, 5-iodocytosine, or N4-methylcytosine; N6-methyladenine or 8-bromoadenine; and N2-methylguanine or 8-bromoguanine. The modified base is preferably 5-methylcytosine.

[0055] Examples of modified internucleoside linkages include, but are not limited to, phosphorothioate linkages, phosphorodithioate linkages, phosphotriester linkages, methylphosphonate linkages, methylthiophosphonate linkages, boranophosphate linkages, phosphoramidate linkages, and the like.

[0056] Examples of the modification of the sugar moiety include, for example, 2'-O-methoxyethyl modification of the sugar moiety, 2'-O-methyl modification of the sugar moiety, 2'-fluoro modification of the sugar moiety, crosslinking between the 2' position and the 4' position of the sugar moiety (the nucleotide having such a crosslinked structure is a crosslinked nucleotide), and the like. Examples of the crosslinked nucleotide include, for example, locked nucleic acid (LNA), 2'-O,4'-C-ethylenebridged nucleic acid (ENA), and the like. More specifically, examples of the crosslinked nucleotide include those having the following nucleoside structure.

[0057] [Chemical formula]

[0058] (In the formula, R represents a hydrogen atom, an alkyl group having 1 to 7 carbon atoms which may form a branch or a ring, an alkenyl group having 2 to 7 carbon atoms which may form a branch or a ring, an aryl group having 3 to 12 carbon atoms which may contain a hetero atom, an aralkyl group having an aryl moiety having 3 to 12 carbon atoms which may contain a hetero atom, or a protecting group for the amino group in nucleic acid synthesis. Preferably, R is a hydrogen atom, a methyl group, an ethyl group, an n-propyl group, an isopropyl group, a phenyl group, or a benzyl group, and more preferably, R is a hydrogen atom or a methyl group. Base is a natural base or a modified base.)

[0059] Further, the modified nucleotide may be PNA, UNA (unlocked nucleic acid), HNA, morpholino nucleic acid, or the like. In the above HNA, the hydroxyl group in the hexopyranose moiety may be deoxygenated. Further, in the above HNA, the hydroxyl group in the hexopyranose moiety may be substituted with a fluorine atom.

[0060] The nucleic acid of the present invention may be modified at the 5'-end, 3'-end and / or inside the nucleic acid strand (in the case of double-stranded nucleic acid, each nucleic acid strand) with one or more ligands or fluorophores. The nucleic acid modified with a ligand or fluorophore is also called a conjugate nucleic acid. During the extension reaction on a solid phase, a modifier capable of reacting on the solid phase can be reacted to modify the 5'-end, 3'-end and / or inside the sequence. Also, nucleic acids introduced with functional groups such as amino group, mercapto group, azide group or triple bond can be synthesized and purified in advance, and conjugate nucleic acids can be obtained by allowing a modifying agent to act on them. The ligand may be any molecule having an affinity for a biomolecule. For example, lipids such as cholesterol, fatty acid, tocopherol, retinoid, saccharides such as N-acetylgalactosamine (GalNAc), galactose (Gal), mannose (Man), antibodies such as full antibody, Fab, VHH, proteins such as low density lipoprotein (LDL), human serum albumin, peptides such as RGD, NGR, R9, CPP, small molecules such as folic acid, synthetic polymers such as synthetic polyamino acids, or nucleic acid aptamers, etc. can be mentioned, and these can also be used in combination. Examples of the fluorophore include Cy3 series, Alexa series, black hole quencher, etc.

[0061] Administration of the medicament of the present invention to a mammal can be carried out orally or parenterally (for example, subcutaneous injection, intramuscular injection, local injection (e.g., intracerebroventricular administration, intrathecal administration), intraperitoneal administration, etc.), but it is desirable to administer it parenterally (particularly, by intracerebroventricular administration or intrathecal administration).

[0062] Compositions for oral administration include solid or liquid dosage forms, specifically tablets (including sugar-coated tablets and film-coated tablets), pills, granules, powders, capsules (including soft capsules), syrups, emulsions, suspensions, etc. On the other hand, as compositions for parenteral administration, for example, injections, suppositories, etc. are used, and injections may include dosage forms such as intravenous injections, subcutaneous injections, intradermal injections, intramuscular injections, drip injections, etc. These preparations contain excipients (such as sugar derivatives like lactose, sucrose, glucose, mannitol, sorbitol; starch derivatives like corn starch, potato starch, α-starch, dextrin; cellulose derivatives like crystalline cellulose; gum arabic; dextran; organic excipients like pullulan; and silicate derivatives like light anhydrous silicic acid, synthetic aluminum silicate, calcium silicate, magnesium aluminometasilicate; phosphates like calcium hydrogen phosphate; carbonates like calcium carbonate; sulfates like calcium sulfate, etc., which are inorganic excipients), lubricants (such as metal stearates like stearic acid, calcium stearate, magnesium stearate; talc; colloidal silica; waxes like beeswax, gay wax; boric acid; adipic acid; sulfates like sodium sulfate; glycols; fumaric acid; sodium benzoate; DL-leucine; lauryl sulfates like sodium lauryl sulfate, magnesium lauryl sulfate; silicic acids like anhydrous silicic acid, hydrated silicic acid; and the above-mentioned starch derivatives), binders (such as hydroxypropyl cellulose, hydroxypropyl methylcellulose, polyvinyl pyrrolidone, macrogol, and compounds similar to the above excipients), disintegrants (such as cellulose derivatives like low-substituted hydroxypropyl cellulose, carboxymethyl cellulose, carboxymethyl cellulose calcium, sodium carboxymethyl cellulose crosslinked internally; chemically modified starch-celluloses like carboxymethyl starch, sodium carboxymethyl starch, crosslinked polyvinyl pyrrolidone), emulsifiers (such as colloidal clays like bentonite, veegum; metal hydroxides like magnesium hydroxide, aluminum hydroxide; anionic surfactants like sodium lauryl sulfate, calcium stearate;Cationic surfactants such as benzalkonium chloride; and nonionic surfactants such as polyoxyethylene alkyl ethers, polyoxyethylene sorbitan fatty acid esters, and sucrose fatty acid esters), stabilizers (parabens such as methylparaben and propylparaben; alcohols such as chlorobutanol, benzyl alcohol, and phenylethyl alcohol; benzalkonium chloride; phenols such as phenol and cresol; thimerosal; dehydroacetic acid; and sorbic acid), flavoring and odor-correcting agents (e.g., sweeteners, acidulants, fragrances, etc. commonly used), diluents, and other additives are used to produce it by a well-known method.;

[0063] Examples of pharmaceutically acceptable carriers include, but are not limited to, excipients such as sucrose and starch, binders such as cellulose and methylcellulose, disintegrants such as starch and carboxymethylcellulose, lubricants such as magnesium stearate and aerosil, fragrances such as citric acid and menthol, preservatives such as sodium benzoate and sodium bisulfite, stabilizers such as citric acid and sodium citrate, suspending agents such as methylcellulose and polyvinylpyrrolidone, dispersants such as surfactants, diluents such as water and physiological saline, and base waxes etc.

[0064] When the KLHL32 gene expression inhibitor is in the form of a nucleic acid, the medicament of the present invention may further contain a nucleic acid transfection reagent to facilitate the introduction of the nucleic acid into target cells. Examples of the nucleic acid transfection reagent include, for example, calcium chloride, Calcium enrichment reagent, atelocollagen; liposomes; nanoparticles; cationic lipids such as lipofectin, lipofectamine, DOGS (transfectamine), DOPE, DOTAP, DDAB, DHDEAB, HDEAB, polybrene, or poly(ethyleneimine) (PEI), etc. can be used.

[0065] Alternatively, the medicament of the present invention may be a pharmaceutical composition in which an agent for suppressing the expression of the KLHL32 gene is encapsulated in liposomes. Liposomes are fine closed vesicles having an inner phase surrounded by one or more lipid bilayers, and can usually hold a water-soluble substance in the inner phase and a lipid-soluble substance in the lipid bilayer. In the present specification, when referring to "encapsulation", the agent for suppressing the expression of the KLHL32 gene may be retained in the inner phase of the liposome or in the lipid bilayer. The liposomes used in the present invention may be monolayer membranes or multilayer membranes, and the particle size can be appropriately selected, for example, in the range of 10 to 1000 nm, preferably 50 to 300 nm. Considering the delivery property to the target tissue, the particle size is, for example, 200 nm or less, preferably 100 nm or less.

[0066] Examples of the method for encapsulating a water-soluble compound such as a nucleic acid into liposomes include the lipid film method (vortex method), reverse evaporation method, surfactant removal method, freeze-thaw method, remote loading method, etc., but are not limited thereto, and any known method can be appropriately selected.

[0067] The dosage of the medicament of the present invention varies depending on the purpose of administration, administration method, type of target disease, severity, and condition of the administration subject (gender, age, weight, etc.). For example, when the agent for suppressing the expression of the KLHL32 gene is systemically administered to an adult in the form of a nucleic acid, usually, the single dose of the expression-suppressing agent is 2 nmol / kg or more and 50 nmol / kg or less, and when administered locally, 1 pmol / kg or more and 10 nmol / kg or less is desirable. It is desirable to administer such a dosage 1 to 10 times, more preferably 5 to 10 times. The dosage may be increased or decreased according to the symptoms.

[0068] The medicament of the present invention can also be used in combination with other prophylactic or therapeutic agents for neurodegenerative diseases (hereinafter sometimes referred to as "existing drugs") (e.g., aducanumab, donepezil, rivastigmine, galantamine, donepezil, memantine, etc.). Therefore, in one aspect of the present invention, there is provided a prophylactic or therapeutic agent for neurodegenerative diseases containing an agent for suppressing the expression of the KLHL32 gene and one or more existing drugs.

[0069] When used as a concomitant drug, such a concomitant drug can be formulated together with an agent for suppressing the expression of the KLHL32 gene and administered as a single preparation, or can be formulated separately from the agent for suppressing the expression of the KLHL32 gene (for example, as a kit) and administered simultaneously or at different times via the same or different routes as the medicament of the present invention. Further, the dosage of these concomitant drugs may be the amount usually used when the drug is administered alone, or may be reduced from the usually used amount.

[0070] 2. A method for screening a preventive or therapeutic agent for neurodegenerative diseases As described above, since neurodegenerative diseases can be prevented or treated by suppressing the expression of the KLHL32 gene, if a substance that suppresses the expression of the KLHL32 gene can be screened, the substance can be used as a candidate for a prophylactic or therapeutic agent for neurodegenerative diseases (in other words, a substance having prophylactic or therapeutic activity against neurodegenerative diseases). Therefore, in another aspect, the present invention provides a method for screening a prophylactic or therapeutic agent for neurodegenerative diseases (hereinafter sometimes referred to as "the screening method 1 of the present invention") including the following steps. (1) A step of bringing a test substance into contact with cells expressing the KLHL32 gene, and (2) A step of selecting a test substance with a reduced expression level of the KLHL32 gene as a candidate for a prophylactic or therapeutic agent for neurodegenerative diseases.

[0071] Further, since the KLHL32 gene is expressed or the expression level of the gene increases prior to neurodegeneration, if a substance that does not increase the expression level of the KLHL32 gene can be screened in a model cell of neurodegeneration, the substance can be used as a candidate for a prophylactic or therapeutic agent for neurodegenerative diseases. Therefore, in yet another aspect, the present invention provides a method for screening a prophylactic or therapeutic agent for neurodegenerative diseases (hereinafter sometimes referred to as "the screening method 2 of the present invention") including the following steps. (1’) A step of bringing a test substance into contact with cells, and Step of selecting a test substance that suppresses the increased expression level of the KLHL32 gene as a candidate for a prophylactic or therapeutic agent for neurodegenerative diseases. Hereinafter, the term "screening method of the present invention" may be used as encompassing Screening Method 1 of the present invention and Screening Method 2 of the present invention.

[0072] The cells to be contacted with the test substance are not particularly limited as long as they are cells that express the KLHL32 gene or cells that will express the KLHL32 gene by culturing or the like, and may be cultured cells, cells isolated from mammals (e.g., humans, mice, rats, hamsters, rabbits, cats, dogs, cows, sheep, monkeys, etc.), or cells within mammals. Specifically, examples of the cells used in the present invention include (1) tissue stem cells (somatic stem cells) such as neural stem cells, hematopoietic stem cells, mesenchymal stem cells, dental pulp stem cells, (2) tissue progenitor cells, and (3) differentiated cells such as nerve cells, lymphocytes, epithelial cells, endothelial cells, muscle cells, fibroblasts (such as skin cells), hair cells, hepatocytes, gastric mucosal cells, intestinal cells, spleen cells, pancreatic cells (such as pancreatic exocrine cells), brain cells, lung cells, kidney cells, and fat cells. Among them, nerve cells are preferred, and human nerve cells are more preferred. In one aspect, the nerve cells used in the present invention are cells that express at least one nerve cell-specific marker gene selected from the group consisting of β-III tubulin, NeuN, N-CAM (neural cell adhesion molecule), and MAP2 (microtubule-associated protein 2), and have β-III tubulin-positive protrusions (i.e., neurites).

[0073] In another aspect, examples of the cells include model cells for neurodegenerative diseases. Examples of the model cells for neurodegenerative diseases include cells derived from patients with neurodegenerative diseases, cells derived from model animals for neurodegenerative diseases, and cells induced to differentiate from pluripotent stem cells having a mutation in the gene that causes the neurodegenerative disease. The pluripotent stem cells having a mutation in the gene that causes the neurodegenerative disease are preferably induced pluripotent stem cells (iPS cells) established by a method known per se using somatic cells collected from a patient with a neurodegenerative disease. Pluripotent stem cells can be differentiated into any cell (preferably, nerve cells) by a method known per se. For specific examples of neurodegenerative diseases, genes that cause neurodegenerative diseases, and variants of the genes, the content described in the above "1. Preventive or therapeutic agents for neurodegenerative diseases" is incorporated herein by reference.

[0074] iPS cells can be produced by introducing specific reprogramming factors into somatic cells in the form of DNA or proteins. Examples of genes included in the reprogramming factors are, for example, Oct3 / 4, Sox2, Sox1, Sox3, Sox15, Sox17, Klf4, Klf2, c-Myc, N-Myc, L-Myc, Nanog, Lin28, Fbx15, ERas, ECAT15-2, Tcl1, beta-catenin, Lin28b, Sall1, Sall4, Esrrb, Nr5a2, Tbx3 or Glis1, etc. These reprogramming factors can be used alone or in combination. Examples of combinations of reprogramming factors are WO2007 / 069666, WO2008 / 118820, WO2009 / 007852, WO2009 / 032194, WO2009 / 058413, WO2009 / 057831, WO2009 / 075119, WO2009 / 079007, WO2009 / 091659, WO2009 / 101084, WO2009 / 101407, WO2009 / 102983, WO2009 / 114949, WO2009 / 117439, WO2009 / 126250, WO2009 / 126251, WO2009 / 126655, WO2009 / 157593, WO2010 / 009015, WO2010 / 033906, WO2010 / 033920, WO2010 / 042800, WO2010 / 050626, WO2010 / 056831, WO2010 / 068955, WO2010 / 098419, WO2010 / 102267, WO2010 / 111409, WO2010 / 111422, WO2010 / 115050, WO2010 / 124290, WO2010 / 147395, WO2010 / 147612, Huangfu D, et al. (2008), Nat. Biotechnol., 26: 795-797, Shi Y, et al. (2008), Cell Stem Cell, 2: 525-528, Eminli S, et al. (2008), Stem Cells. 26:2467-2474, Huangfu D, et al. (2008), Nat Biotechnol. 26:1269-1275, Shi Y, et al.(2008), Cell Stem Cell, 3, 568-574, Zhao Y, et al. (2008), Cell Stem Cell, 3:475-479, Marson A, (2008), Cell Stem Cell, 3, 132-135, Feng B, et al. (2009), Nat Cell Biol. 11:197-203, R.L. Judson et al., (2009), Nat. Biotech., 27:459-461, Lyssiotis CA, et al. (2009), Proc Natl Acad Sci USA. 106:8912-8917, Kim JB, et al. (2009), Nature. 461:649-643, Ichida JK, et al. (2009), Cell Stem Cell. 5:491-503, Heng JC, et al. (2010), Cell Stem Cell. 6:167-74, Han J, et al. (2010), Nature. 463:1096-100, Mali P, et al. (2010), Stem Cells. 28:713-720, Maekawa M, et al. (2011), Nature. 474:225-9. The combinations described therein are exemplified.

[0075] As methods for inducing differentiation of pluripotent stem cells such as iPS cells into specific cells, various known differentiation induction methods can be appropriately selected and used. For example, as a method for differentiating into nerve cells, (1) a method of differentiating into motor nerve cells by expressing three types of (motor) nerve cell lineage-specific transcription factors (Ngn2, Lhx3, and Isl1) in neural progenitor cells prepared by the SFEB method from human-derived embryonic stem cells or induced pluripotent stem cells, as described in Hester M.E. et al., Mol. Therapy, 19:1905-1912 (2011); (2) a method of differentiating into motor nerve cells by expressing the above three types of genes in pluripotent stem cells, as described in WO 2014 / 148646; (3) a method of differentiating into glutamatergic nerve cells in the cerebral cortex by expressing Ngn2 in pluripotent stem cells, as described in WO 2014 / 148646; (4) a method of differentiating into nerve cells by expressing three types of nerve cell lineage-specific transcription factors (Ascl1, Brn2, and Mytl1) in human embryonic stem cells, as described in Pang Z.P. et al., Nature, 476:220-223 (2012), etc. can be mentioned, but it is not limited thereto.

[0076] In the case of Alzheimer's disease model animals, examples include animals in which mutations have been introduced into one or more genes selected from the group consisting of APP, PSEN1, and PSEN2, or variants of these genes have been introduced from the outside. More specifically, (1) PSEN1 transgenic mice (mice expressing an exon 9 deletion mutant (PSEN1dE9) under the control of the mouse PrP promoter, Jankowsky JL et al., Hum Mol Genet. 13(2):159-70 (2004)), (2) PSEN2 transgenic mice (mice expressing a human PSEN2 mutant (N141I) under the control of the ubiquitous CMV early enhancer and the chicken β-actin promoter, Oyama F et al., J. Neurochem. 71: 313-322 (1998)), (3) human double mutant APP695 (KM670 / 671NL, Swedish type) transgenic mice (Hsiao K et al., Science, 274(5284):99-102 (1996)), (4) 5×FAD mice (transgenic mice in which human APP695 with triple mutations of the Swedish type (KM670 / 671NL), Florida type (I716V), and London type (V717I), and human PSEN1 with double mutations (M146L and L285V) are expressed under the control of the mouse Thy1) (Oakley H et al., J. Neurosci. 26, 10129-10140 (2006)), (5) transgenic mice expressing a human tau mutant protein under the control of the mouse PrP promoter (Yoshiyama Y et al., Neuron. 53(3):337-351 (2007)), etc.

[0077] In the case of an animal model of amyotrophic lateral sclerosis, examples include an animal in which one or more genes selected from the group consisting of SOD1, C9ORF72, TDP43, FUS, PRN1, EPH4N, ANG, UBQLN, and HNPNPA have been mutated, or a mutant of these genes has been introduced from the outside. More specifically, transgenic mice obtained by introducing a mutant SOD1 (e.g., one or more mutations selected from the group consisting of A4V, G37R, G41D, H46R, G85R, D90A, G93A, G93S, I112T, I113T, L114F, and S134N) can be mentioned. In addition, examples include a tauopathy model animal in which a mutation has been introduced into the MAPT gene, or a mutant of this gene has been introduced from the outside. More specifically, tauopathy model mice into which one or more mutant MAPT genes selected from the group consisting of G272V, N297K, P301L, P301S, V337M, and R406W have been introduced can be mentioned. The model animals listed here are examples, and other neurodegenerative models can also be used in the screening method of the present invention.

[0078] In the screening method of the present invention, the expression level of the KLHL32 gene (that is, the expression level of the KLHL32 transcript or the expression level of the KLHL32 protein) can be used as an index. When the expression level serving as these indices is lower than the expression level in cells before contact with the test substance, the expression level in cells not in contact with the test substance, or the expression level in cells contacted with a substance (control substance) known not to show a preventive or therapeutic effect on neurodegenerative diseases, it can be evaluated that the test substance has decreased the expression level of the KLHL32 gene, or that the test substance has suppressed an increase in the expression level of the KLHL32 gene.

[0079] The expression level of the KLHL32 transcript can be detected or quantified by known methods using, for example, transcripts extracted from cells. Examples of such methods include the RT-qPCR method and the digital PCR method. Specifically, RT-qPCR can be performed by a well-known method. For example, cDNA is synthesized from total RNA as a template using reverse transcriptase, and PCR is carried out in the presence of a set of primers specific to the target gene, DNA polymerase, and a dye or probe (e.g., TaqMan (registered trademark) probe, etc.) that can function as a DNA intercalator for quantifying the expression level.

[0080] The primers used for RT-qPCR are present on the synthesized cDNA and can be a set of primers specific to the gene to be amplified. The primers can be designed considering factors such as the amplification size (e.g., preferably 80 - 150 bp), primer size (e.g., 17 - 25 bases), GC content (e.g., 40 - 60%), the sequence at the 3' end (e.g., make the base at the 3' end G or C as much as possible, avoid primers with too high GC content near the 3' end), the distance between sequences (e.g., avoid sequence repeats), sequence complementarity (e.g., make sure there is no complementarity of 3 bases or more within the upstream primer or between primers), Tm value (e.g., make the Tm values of the upstream and downstream primers the same. The Tm value is 2(A + T) + 4(G + C)), etc., and can also be designed using primer design software well-known to those skilled in the art. The design of PCR primers can also be outsourced to Applied Biosysytems Inc. etc.

[0081] The expression level of the KLHL32 protein can be detected or quantified by known methods. For example, it can be detected or quantified by Western blot, immunostaining, enzyme immunoassay (e.g., EIA, ELISA), etc. using an antibody that specifically recognizes the KLHL32 protein, but is not limited to these methods as long as the expression level of the target KLHL32 protein can be detected or quantified.

[0082] Examples of test substances used in the screening method of the present invention include, for example, cell extracts, cell culture supernatants, microbial fermentation products, extracts derived from marine organisms, plant extracts, purified proteins or crude proteins, peptides, non-peptide compounds, synthetic low-molecular-weight compounds, and natural compounds. The test substances can also be obtained using any of a number of approaches in combinatorial library methods known in the art, including (1) biological libraries, (2) synthetic library methods using deconvolution, (3) the "one-bead one-compound" library method, and (4) synthetic library methods using affinity chromatography selection. The biological library method using affinity chromatography selection is limited to peptide libraries, but the other four approaches can be applied to peptide, non-peptide oligomer, or low-molecular-weight compound libraries of compounds (Lam (1997) Anticancer Drug Des. 12:145-67). Examples of methods for synthesizing molecular libraries can be found in the art (DeWitt et al. (1993) Proc. Natl. Acad. Sci. USA 90:6909-13; Erb et al. (1994) Proc. Natl. Acad. Sci. USA 91:11422-6; Zuckermann et al. (1994) J. Med. Chem. 37:2678-85; Cho et al. (1993) Science 261:1303-5; Carell et al. (1994) Angew. Chem. Int. Ed. Engl. 33:2059; Carell et al. (1994) Angew. Chem. Int. Ed. Engl. 33:2061; Gallop et al. (1994) J. Med. Chem. 37:1233-51).Compound libraries can be prepared as solutions (see Houghten (1992) Bio / Techniques 13:412-21), beads (Lam (1991) Nature 354:82-4), chips (Fodor (1993) Nature 364:555-6), bacteria (U.S. Patent No. 5,223,409), spores (U.S. Patent Nos. 5,571,698, 5,403,484, and 5,223,409), plasmids (Cull et al. (1992) Proc. Natl. Acad. Sci. USA 89:1865-9), or phages (Scott and Smith (1990) Science 249:386-90; Devlin (1990) Science 249:404-6; Cwirla et al. (1990) Proc. Natl. Acad. Sci. USA 87:6378-82; Felici (1991) J. Mol. Biol. 222:301-10; U.S. Patent Application No. 2002103360).

[0083] Contact between the test substance and the cells can be achieved by adding the test substance to the medium or buffer containing the cells in the case of cultured cells, or by orally or parenterally administering the test substance to a non-human mammal in the case of cells in a mammal. The amount added to the medium, etc., and the dosage administered to the mammal can be appropriately selected according to the test substance. Contact between the test substance and the cells is not particularly limited, and examples include 1 day or more, 2 days or more, 3 days or more, 4 days or more, 5 days or more, 6 days or more, 7 days or more. The concentration of the test substance added can be appropriately adjusted according to the type of compound (solubility, toxicity, etc.).

[0084] In the screening method of the present invention, the cell culture medium used when bringing a test substance into contact with cells is not particularly limited as long as it is a medium capable of culturing the cells, and may be a basal medium, but preferably a medium suitable for inducing differentiation into nerve cells (hereinafter sometimes referred to as "nerve differentiation induction medium"). The nerve differentiation induction medium can be prepared by adding a neurotrophic factor to a basal medium supplemented with a neurotrophic factor. A neurotrophic factor is a ligand of a membrane receptor that plays an important role in the survival and function maintenance of nerve cells. Examples thereof include Nerve Growth Factor (NGF), Brain-derived Neurotrophic Factor (BDNF), Neurotrophin 3 (NT-3), Neurotrophin 4 / 5 (NT-4 / 5), Neurotrophin 6 (NT-6), basic FGF, acidic FGF, FGF-5, Epidermal Growth Factor (EGF), Hepatocyte Growth Factor (HGF), Insulin, Insulin Like Growth Factor 1 (IGF 1), Insulin Like Growth Factor 2 (IGF 2), Glia cell line-derived Neurotrophic Factor (GDNF), TGF-b2, TGF-b3, Interleukin 6 (IL-6), Ciliary Neurotrophic Factor (CNTF), and LIF. Among these, preferred neurotrophic factors in the present invention are GDNF, BDNF, and / or NT-3.

[0085] Examples of the basal medium include Glasgow's Minimal Essential Medium (GMEM) medium, IMDM medium, Medium 199 medium, Eagle's Minimum Essential Medium (EMEM) medium, αMEM medium, Dulbecco's modified Eagle's Medium (DMEM) medium, Ham's F12 (F12) medium, Dulbecco's Modified Eagle Medium: Nutrient Mixture F-12 (DMEM / F-12) medium, RPMI 1640 medium, Fischer's medium, Neurobasal Medium (Lifetechnologies), and mixed media thereof. The basal medium may or may not contain serum.

[0086] If necessary, the medium may contain one or more serum substitutes such as Knockout Serum Replacement (KSR) (serum substitute for FBS during ES cell culture), N2 supplement (Invitrogen), B27 supplement (Invitrogen), albumin, transferrin, apotransferrin, fatty acids, insulin, collagen precursors, trace elements, 2-mercaptoethanol, 3'-thiol glycerol, etc. It may also contain one or more substances such as lipids, amino acids, L-glutamine, Glutamax (Invitrogen), non-essential amino acids, vitamins, growth factors, low molecular weight compounds, antibiotics, antioxidants, pyruvic acid, buffers, inorganic salts, selenic acid, progesterone, and putrescine.

[0087] The culture temperature when contacting the test substance with the cells is not particularly limited, but is about 30 to 40 °C, preferably about 37 °C, and the culture is carried out in an atmosphere of air containing CO2, and the CO2 concentration is preferably about 2 to 5%.

[0088] 3. A biomarker for diagnosing neurodegenerative diseases and its use As shown in the following examples, upon analyzing the autopsy brains of AD patients, KLHL32 expression was observed specifically in the sites where β-amyloid accumulates. In an Alzheimer's disease mouse model, it was confirmed that KLHL32 is expressed in neurons prior to the activation of caspase-3. Therefore, KLHL32 protein and KLHL32 transcripts can be used as biomarkers for diagnosing neurodegenerative diseases (hereinafter sometimes referred to as "the biomarkers of the present invention"). Multiple isoforms (i.e., isoforms a to o, X1 to X3) of KLHL32 protein are known. Unless otherwise specified, the term "KLHL32 protein" is used hereinafter to include all isoforms. Among the isoforms of KLHL32 protein, the full-length KLHL32 protein (isoform a in the case of human KLHL32 protein) is preferred as a biomarker for diagnosing neurodegenerative diseases. Therefore, unless otherwise specified, the term "KLHL32 transcript" is used hereinafter to include all isoforms. Among the isoforms of KLHL32 transcript, KLHL32 mRNA encoding the full-length KLHL32 protein is preferred as a biomarker for diagnosing neurodegenerative diseases.

[0089] The KLHL32 protein used as a biomarker of the present invention is a known protein. In the case of humans, the amino acid sequences are disclosed as NCBI Accession No.: NP_443136.2 or NP_443136.2 (isoform a; SEQ ID NO: 3), NP_001273179.1 (isoform b), NP_001273180.1 (isoform c), NP_001273181.1 (isoform d), NP_001273183.1 (isoform e), NP_001310182.1 (isoform f), NP_001310183.1 (isoform g), NP_001310184.1 (isoform h), NP_001310185.1 (isoform i), NP_001310186.1 (isoform j), NP_001310187.1 (isoform k), NP_001310189.1 (isoform l), NP_001310191.1 (isoform m), NP_001310192.1 (isoform n), NP_001310193.1 (isoform o), XP_005266870.1 (isoform X1) and XP_016865717.1 (isoform X2), XP_016865718.1 (isoform X3). In the present invention, the KLHL32 protein may be an amino acid sequence represented by any of the NCBI Accession Nos., a protein containing an amino acid sequence substantially identical to these, or a protein encoded by an ortholog of the human KLHL32 gene in animals other than humans.

[0090] Examples of amino acid sequences that are substantially identical to the amino acid sequence represented by any of the above NCBI Accession Nos. (for example, the sequence shown in SEQ ID NO: 3) include, for example, amino acid sequences having 60% or more, preferably 70% or more, more preferably 80% or more, still more preferably 90% or more, even more preferably 95% or more, and most preferably 98% or more similarity or identity to these amino acid sequences. Here, "similarity" means the percentage of identical and similar amino acid residues to all overlapping amino acid residues in the optimal alignment (preferably, the algorithm can consider the introduction of gaps into one or both of the sequences for the optimal alignment) when two amino acid sequences are aligned using a mathematical algorithm known in the art. "Similar amino acids" mean amino acids that are similar in physicochemical properties, for example, aromatic amino acids (Phe, Trp, Tyr), aliphatic amino acids (Ala, Leu, Ile, Val), polar amino acids (Gln, Asn), basic amino acids (Lys, Arg, His), acidic amino acids (Glu, Asp), amino acids having a hydroxyl group (Ser, Thr), and amino acids classified into the same group such as amino acids with small side chains (Gly, Ala, Ser, Thr, Met). Substitutions with such similar amino acids are predicted not to change the phenotype of the protein (i.e., they are conservative amino acid substitutions). Specific examples of conservative amino acid substitutions are well-known in the art and are described in various documents (see, for example, Bowie et al., Science, 247: 1306-1310 (1990)). The similarity or identity of amino acid sequences in this specification can be calculated using the homology calculation algorithm NCBI BLAST (National Center for Biotechnology Information Basic Local Alignment Search Tool) under the following conditions (expect value = 10; gaps are allowed; matrix = BLOSUM62; filtering = OFF).

[0091] The KLHL32 protein can be produced according to known protein synthesis methods, such as solid-phase synthesis method, liquid-phase synthesis method, etc. The obtained protein can be purified and isolated by known purification methods, such as solvent extraction, distillation, column chromatography, liquid chromatography, recrystallization, combinations thereof, etc. It can also be isolated and purified from biological samples by methods known per se. Alternatively, the KLHL32 protein can also be produced by culturing a transformant containing the nucleic acid encoding it and separating and purifying the protein from the resulting culture. Such nucleic acid may be DNA, RNA, or a DNA / RNA chimera, but is preferably DNA. The nucleic acid may be double-stranded or single-stranded.

[0092] In addition, amino acid sequences substantially identical to the amino acid sequence represented by any of the above NCBI Accession No include proteins containing amino acid sequences in which one or more (preferably about 1 to 100, preferably about 1 to 50, more preferably about 1 to 10, and particularly preferably 1 to several (2, 3, 4, or 5)) amino acids of these amino acid sequences are substituted, inserted, and / or deleted.

[0093] The KLHL32 transcript is a known transcript, for example, RNA encoding the above KLHL32 protein. Specifically, in the case of humans, not only RNA encoding isoform a shown in SEQ ID NO: 1 or 2, but also NM_001286250.2 (RNA encoding isoform b), NM_001286251.2 (RNA encoding isoform c), NM_001286252.2 (RNA encoding isoform d), NM_001286254.3 (RNA encoding isoform e), NM_001323253.2 (RNA encoding isoform f), NM_001323254.2 (RNA encoding isoform g), NM_001323255.2 (RNA encoding isoform h), NM_001323256.2 (RNA encoding isoform i), NM_001323257.2 (RNA encoding isoform j), NM_001323258.2 (RNA encoding isoform k), NM_001323260.2 (RNA encoding isoform l), NM_001323262.2 (RNA encoding isoform m), NM_001323263.2 (RNA encoding isoform n), NM_001323264.2 (RNA encoding isoform o), XM_005266813.5 (RNA encoding isoform X1), XM_017010228.2 (RNA encoding isoform X2), and XM_017010229.2 (RNA encoding isoform X3) have their nucleotide sequences disclosed. In the present invention, the KLHL32 transcript may be any nucleic acid having the above nucleotide sequence (provided that T is read as U), or a nucleic acid containing a nucleotide sequence identical or substantially identical to these nucleotide sequences. Further, it may be a transcript transcribed from an ortholog of the human KLHL32 gene of an animal other than human.

[0094] Examples of the nucleic acid containing a nucleotide sequence substantially identical to the above nucleotide sequence include a nucleic acid containing a nucleotide sequence having 60% or more, preferably 70% or more, more preferably 80% or more, still more preferably 90% or more, even more preferably 95% or more, and most preferably 98% or more identity with these nucleotide sequences and encoding a protein having substantially the same activity as the KLHL32 protein.

[0095] The KLHL32 transcript can be obtained, for example, by isolating and purifying it from a biological sample containing the transcript by a method known per se.

[0096] In yet another aspect, the present invention provides a method for diagnosing (e.g., diagnosing, determining, judging, differentiating, examining) whether a subject has a neurodegenerative disease, a method for assisting such diagnosis, or a method for analyzing (e.g., analyzing, evaluating, calculating, etc.) the possibility or probability of a neurodegenerative disease, which includes the step of detecting a biomarker of the present invention in a subject or a sample derived from the subject (hereinafter, these may be collectively referred to as "the method of the present invention"). The biomarker to be detected may be one kind or two or more kinds (for example, detecting both the KLHL32 protein and the KLHL32 mRNA). As used herein, "assisting diagnosis etc." means providing information serving as an index for determining whether a subject has a neurodegenerative disease, and does not include the step of diagnosing etc. whether the subject has a neurodegenerative disease, which is a medical act. Also, as used herein, "detecting a biomarker" includes not only examining whether the expression of the KLHL32 gene is observed (i.e., whether there is a protein or transcript in an amount above the detection limit in the detection method) in at least a part of the subject's brain (e.g., senile plaques, degenerated neurites, etc.) or a sample derived from the subject, but also measuring (quantifying) the expression level thereof.

[0097] The subject of the method of the present invention may be an animal other than a human. Preferably, animals suspected of having a neurodegenerative disease are included. Examples of the type of animal include mammals (e.g., humans, monkeys, cows, pigs, horses, dogs, cats, sheep, goats, rabbits, hamsters, guinea pigs, mice, rats, etc.), birds (e.g., chickens, etc.). Preferably, they are mammals, more preferably humans.

[0098] Samples derived from a subject include, for example, blood, serum, plasma, saliva, urine, tears, sweat, milk, nasal mucus, semen, pleural effusion, gastrointestinal secretions, cerebrospinal fluid, interstitial fluid, and lymph fluid. From a clinical perspective, blood, serum, plasma, and cerebrospinal fluid are preferred, and serum and plasma are more preferred. These samples can be obtained by methods known per se. For example, serum and plasma can be prepared by collecting blood from a subject according to a conventional method and separating the liquid components, and cerebrospinal fluid can be collected by known means such as lumbar puncture.

[0099] The detection of KLHL32 transcripts in a sample derived from a subject can be examined by preparing an RNA (e.g., total RNA, mRNA) fraction from the sample and detecting the KLHL32 transcripts contained in the fraction. Therefore, in one embodiment, the method of the present invention includes detecting using a nucleic acid probe or nucleic acid primer that can specifically recognize KLHL32 transcripts, respectively.

[0100] The preparation of the RNA fraction can be carried out using known methods such as guanidine-CsCl ultracentrifugation method and AGPC method, and it is also possible to rapidly and simply prepare high-purity total RNA from a small amount of sample using a commercially available RNA extraction kit (e.g., RNeasy Mini Kit; manufactured by QIAGEN, etc.). As a means for detecting the KLHL32 transcript in the RNA fraction, for example, methods using hybridization (Northern blot, dot blot, etc.), or methods using PCR (RT-PCR, competitive PCR, real-time PCR, etc.) can be mentioned. Quantitative PCR methods such as competitive PCR and real-time PCR are preferable in that they can rapidly and simply detect the expression from a small sample.

[0101] In the case of Northern blot or dot blot hybridization, the detection of the KLHL32 transcript can be carried out, for example, using a nucleic acid probe that can specifically recognize the KLHL32 transcript. The nucleic acid probe may be DNA, RNA, or a DNA / RNA chimera, but is preferably DNA. Also, the nucleic acid used as the probe may be double-stranded or single-stranded. In the case of double-stranded, it may be double-stranded DNA, double-stranded RNA, or a DNA:RNA hybrid. In the case of single-stranded, those containing an antisense strand sequence can be used.

[0102] The above nucleic acid probe can be obtained, for example, by chemically synthesizing it using a commercially available DNA / RNA automatic synthesizer, etc. Also, the above nucleic acid probe is preferably labeled with a labeling agent in order to enable the detection of the target nucleic acid.

[0103] The detection of KLHL32 protein in a sample derived from a subject can be examined by preparing a protein fraction from the sample and detecting or quantifying the translation product of the gene (i.e., KLHL32 protein) contained in the fraction. The detection or quantification of these proteins can be carried out by immunological assays (e.g., ELISA, FIA, RIA, Western blot, etc.) using antibodies that specifically recognize each protein, but preferably, an immunological assay, particularly a method using ELISA, is more preferred.

[0104] Antibodies that can specifically recognize KLHL32 protein can be produced by using these proteins or partial peptides having epitopes as immunogens and by existing common production methods. In the present specification, antibodies include, but are not limited to, natural antibodies such as polyclonal antibodies and monoclonal antibodies (mAbs), chimeric antibodies, humanized antibodies, single-chain antibodies, and binding fragments thereof that can be produced using genetic recombination techniques. Preferably, the antibody is a polyclonal antibody, a monoclonal antibody, or a binding fragment thereof. The binding fragment means a region of a part of the aforementioned antibody having specific binding activity, and specifically, for example, F(ab’)2, Fab’, Fab, Fv, sFv, dsFv, sdAb, etc. can be mentioned (Exp. Opin. Ther. Patents, Vol.6, No.5, p.441-456, 1996). The class of the antibody is not particularly limited and includes antibodies having any isotype such as IgG, IgM, IgA, IgD, or IgE. Preferably, it is IgG or IgM, and more preferably IgG in consideration of ease of purification, etc. Also, in the present invention, it is also preferable to use commercially available antibodies or kits or arrays containing antibodies as antibodies that can specifically recognize KLHL32 protein.

[0105] The detection of biomarkers in a subject can also be performed, for example, using an antibody directly or indirectly labeled with a radionuclide. Such radionuclides include nuclides that emit positrons (e.g., 68 Ga, 64 Cu, 86 Y, 89 Zr, etc.) that can be used in PET (Positron Emission Tomography) examinations, and nuclides that emit gamma rays (e.g., 111 In, etc.) that can be used in SPECT (Single Photon Emission Computed Tomography) examinations. The preparation of an antibody labeled with a radionuclide and the detection of a biomarker using the antibody can be carried out by methods known per se (e.g., the methods described in WO2013 / 157102, WO2021 / 075544, etc.).

[0106] If necessary, the above antibody may be encapsulated in microcapsules (such as microcapsules of hydroxymethylcellulose, gelatin, methyl polymethacrylate, etc.) and may be in the form of a colloidal drug delivery system (such as liposomes, albumin microspheres, microemulsions, nanoparticles, and nanocapsules, etc.) ("Remington’s Pharmaceutical Science 16 th edition", Oslo Ed. (1980), etc.). Further, a functional peptide such as a cell membrane permeable peptide (e.g., TAT peptide, a modified form of the peptide, etc.) may be added to the antibody.

[0107] In addition, by comparing the amounts of the biomarkers of the present invention, it is also possible to perform a diagnosis or the like of a neurodegenerative disease. Specifically, for example, it can be performed by quantifying the amount of the biomarker of the present invention in a sample from a healthy subject or a healthy subject (hereinafter sometimes referred to as a "control sample") and a sample from a subject or a subject of interest of the present invention, and comparing the amounts of both. Alternatively, it may be performed by comparing the amount of the biomarker with a reference value. As the "reference value" used in the present invention, the amount of the biomarker of the present invention in the control sample may be used. Alternatively, a value set in advance from the quantified value of the biomarker in the control sample may be used. In this case, as the reference value, for example, the average value or the mode value of the measured values of a plurality of individuals can be adopted by using a plurality of individuals as a control group. In the present specification, a "healthy subject" means a person who has not been definitely diagnosed with a neurodegenerative disease.

[0108] The above reference value may be a cut-off value. The "cut-off value" is a value that can satisfy both high diagnostic sensitivity (true positive rate) and high diagnostic specificity (true negative rate) when a disease is judged based on that value. For example, a value that shows a high positive rate in a group of patients with neurodegenerative diseases and a high negative rate in a group of healthy subjects can be set as the cut-off value.

[0109] The method for calculating the cut-off value is well known in this field. For example, the amount of the biomarker of the present invention in serum collected from a group of patients with neurodegenerative diseases and a group of healthy subjects is quantified, the diagnostic sensitivity and diagnostic specificity in the quantified values are determined, and based on these values, an ROC (Receiver Operating Characteristic) curve is created using commercially available analysis software. Then, a value when the diagnostic sensitivity and the diagnostic specificity are as close to 100% as possible is determined, and that value can be used as the cut-off value.

[0110] As a result of comparing the amounts of the biomarker of the present invention, for example, if the biomarker of the present invention is at a higher level or equal to or higher than the above reference value in a subject or a sample derived from the subject, compared to a healthy subject or a control sample, the subject can be diagnosed as having a neurodegenerative disease. If the biomarker of the present invention is at the same level or lower or less than the above reference value, compared to a healthy subject or a control sample, the subject can be diagnosed as not having a neurodegenerative disease.

[0111] Also, as a result of comparing the amounts of the biomarker, for example, if the biomarker of the present invention is detected in a subject or a sample derived from the subject, or is at a higher level or equal to or higher than the above reference value, compared to a healthy subject or a control sample, it can be analyzed that the possibility or probability that the subject has a neurodegenerative disease is high. If it is at the same level or lower or less than the above reference value, compared to a healthy subject or a control sample, it can be analyzed that the possibility or probability that the subject has a neurodegenerative disease is low.

[0112] As a result of the method of the present invention, when a subject is diagnosed as having a neurodegenerative disease or the like, based on the result of the diagnosis or the like, a prophylactic or therapeutic agent for neurodegenerative disease to be administered to the subject is selected or determined, and by administering a therapeutically effective amount of the prophylactic or therapeutic agent to the subject, the neurodegenerative disease can be prevented or treated. That is, a method for preventing or treating a neurodegenerative disease including the following steps (i) to (iii) is also provided. (i) A step of detecting the biomarker of the present invention in a subject or a sample derived from the subject; (ii) A step of diagnosing whether the subject has a neurodegenerative disease based on the result of the step (i); (iii) A step of administering a prophylactic or therapeutic agent for neurodegenerative disease to a subject diagnosed as having a neurodegenerative disease in (ii).

[0113] Examples of the prophylactic or therapeutic agent for neurodegenerative disease include the medicament of the present invention, the existing medicaments described in 1., combinations thereof, and the like.

[0114] The above prophylactic or therapeutic agent may be administered orally or parenterally as an active ingredient alone or in admixture with a pharmaceutically acceptable carrier, excipient, diluent or the like, as a pharmaceutical composition in an appropriate dosage form. Examples of compositions for oral administration include solid or liquid dosage forms, specifically tablets (including sugar-coated tablets and film-coated tablets), pills, granules, powders, capsules (including soft capsules), syrups, emulsions, suspensions and the like. On the other hand, as compositions for parenteral administration, for example, injections, suppositories and the like are used, and injections may include dosage forms such as intravenous injections, subcutaneous injections, intradermal injections, intramuscular injections, drip infusions and the like. Further, the dosage of the prophylactic or therapeutic agent can be appropriately set according to various conditions such as the type of the compound, the symptoms of the administration subject, age, body weight, drug tolerance and the like.

[0115] 4. A diagnostic kit for neurodegenerative diseases Furthermore, the present invention provides a kit for diagnosing neurodegenerative diseases (hereinafter sometimes referred to as "the diagnostic kit of the present invention") comprising an antibody that specifically recognizes KLHL32 protein or a nucleic acid probe or nucleic acid primer that specifically recognizes KLHL32 transcript. Regarding the definition, specific examples, detection methods and the like of KLHL32 protein, an antibody that specifically recognizes the protein, KLHL32 transcript, a nucleic acid probe or nucleic acid primer that specifically recognizes the transcript, the contents described in the above "3. Biomarkers for diagnosing neurodegenerative diseases and their uses" are incorporated herein by reference.

[0116] In one aspect, the diagnostic kit of the present invention is also provided as a kit (hereinafter, also referred to as "the simple kit of the present invention") that can detect the presence or absence of the biomarker of the present invention in a sample simply by bringing the sample into contact with a substrate using the ELISA method. The simple kit of the present invention includes a substrate on which an antibody that specifically recognizes the KLHL32 protein (hereinafter, also referred to as "the first antibody") is immobilized. Further, the simple kit of the present invention preferably contains an antibody that specifically recognizes the KLHL32 protein (hereinafter, also referred to as "the second antibody") (typically, the antibody is a labeled antibody) different from the above antibody. With such an aspect, the biomarker of the present invention can be detected by the sandwich ELISA method.

[0117] When the determination kit of the present invention contains the above nucleic acid probe or nucleic acid primer (simply referred to as "nucleic acid") as a component, these nucleic acids can be provided as a solid in a dried state or an alcohol-precipitated state, or can be provided in a state dissolved in water or an appropriate buffer (e.g., TE buffer, etc.). When used as a labeled probe, the nucleic acid can be provided in a state pre-labeled with any of the above labeling substances, or can be provided separately from the labeling substances and labeled at the time of use. Alternatively, the nucleic acid can be provided in a state immobilized (also referred to as supported or solid-phased) on an appropriate substrate. Examples of the substrate include, but are not limited to, glass, silicon, plastic, nitrocellulose, nylon, polyvinylidene difluoride, etc. Also, as the immobilization means, functional groups such as amino groups, aldehyde groups, SH groups, and biotin are introduced into the nucleic acid in advance, while functional groups (e.g., aldehyde groups, amino groups, SH groups, streptavidin, etc.) that can react with the nucleic acid are also introduced onto the substrate, and the substrate and the nucleic acid are cross-linked by a covalent bond between the two functional groups, or for polyanionic nucleic acids, the substrate is coated with polycation to immobilize the nucleic acid using electrostatic binding. However, the methods are not limited to these.

[0118] In addition to the above nucleic acids and antibodies, the diagnostic kit of the present invention may contain other substances necessary for the reaction to detect the expression of the biomarker of the present invention. These other substances may be provided in a coexisting state with nucleic acids, antibodies, etc., as long as they do not adversely affect the reaction, or may be provided together with separate reagents. For example, when the reaction for detecting the expression of the biomarker of the present invention is PCR, examples of such other substances include reaction buffers, dNTPs, heat-resistant DNA polymerases, etc. When competitive PCR or real-time PCR is used, competitor nucleic acids, fluorescent reagents (such as the above intercalators and fluorescent probes), etc. can be further included. When the reaction for detecting the expression of the biomarker of the present invention is an antigen-antibody reaction, examples of such other substances include reaction buffers, competitor antibodies, labeled secondary antibodies (for example, when the primary antibody is a rabbit antibody, mouse anti-rabbit IgG labeled with peroxidase, alkaline phosphatase, etc.), blocking solutions, ELISA plates, etc. In addition, the determination kit of the present invention may contain an instruction manual describing the usage method of the kit and reagents, the determination criteria for diseases, etc. Further, the above determination kit may contain one or more biomarkers of the present invention for use as a positive control, for example.

[0119] The present invention will be described more specifically with reference to the following examples, but the present invention is not limited thereto.

Examples

[0120] <Materials and Methods> (Ethics) The collection of all human tissues, human stem cell research, procedures, and written consent were approved by the Ethics Committee of the Department of Medicine and Graduate School of Medicine, Kyoto University.

[0121] (Establishment of human iPSCs) In the case of iPSCs (induced pluripotent stem cells) derived from human peripheral blood mononuclear cells (PBMCs), human cDNA for reprogramming factors was transduced into human PBMCs using an episomal vector (SOX2, KLF4, OCT4, L-MYC, LIN28, p53 dominant negative). Several days after transduction, PBMCs were collected and re-seeded onto dishes coated with iMatrix. The next day, the medium was changed to StemFit AK03 or AK02N. Thereafter, the medium was changed every other day. Twenty days after transduction, iPSC colonies were picked up. The established PBMC-derived iPSCs were expanded in culture (Kondo T. et al., Cell Rep. 21, 2304-2312 (2017)).

[0122] (Generation of iN-iPSCs) To establish a robust and rapid differentiation method, direct conversion technology was utilized. Human neurogenin 2 (NGN2) cDNA under the tetracycline-inducible promoter (tetO) was transfected into iPSCs using the piggyBac transposon system and Lipofectamine LTX (Thermo Fisher Scientific Inc., Waltham, Massachusetts). A piggyBac vector containing tetO::NGN2 was used (Kim et al., 2016). After antibiotic selection with G418 disulfate (Nacalai-Tesque, Kyoto, Japan), colonies were selected, and subclones that could be efficiently differentiated into neurons by inducing transient expression of NGN2 with MAP2 / DAPI (purity less than 96%) were selected (Kondo T. et al., Cell Rep. 21, 2304-2312 (2017)).

[0123] (Live imaging detection of activated caspase-3) Activated caspase-3 in apoptotic neurons was detected using a fluorogenic substrate consisting of the 4-amino acid peptide DEVD conjugated with a nucleic acid-binding dye (CellEvent Caspase-3 / 7 Green Detection Reagent, Thermo Fisher). Since the DEVD peptide inhibits the ability of the dye to bind to DNA, the reagent is inherently non-fluorescent. However, when activated caspase-3 in dead cells cleaves the DEVD peptide, the fluorescent dye can bind to DNA, generating a fluorescence response. This response can be detected using the standard FITC-green filter set of a conventional time-lapse imaging system (exposure time = 400 ms, Incucyte ZOOM, Essen BioScience, Sartorius group). Images were captured chronically every 6 hours after drug treatment.

[0124] (Data preparation) Two images were captured from the wells of a 384-well plate using Incucyte ZOOM. The captured images were in grayscale and the original size was 1,392 x 1,040 pixels. The original image was evenly divided into four regions, and a square image was cropped from the center of each region. The width x height size of the cropped images was selected from 128 x 128, 256 x 256, 512 x 512, 768 x 768, 1,024 x 1,024. Since the size of the original image was 1,392 x 1,038, when the cropping size was larger than 519 x 519, the cropping area extended beyond the original image. In that case, the cropping area was shifted to fit the edge of the original image. Next, the size of the cropped images was changed to 256 x 256. Images of size 256 x 256 for training were randomly flipped horizontally with a 50% probability, and a region of 224 x 224 pixels was randomly cropped from the image. The cropped region of 224 x 224 was finally used as the input to the network. For the validation data and test data, images of size 224 x 224 were cropped without flipping.

[0125] The logarithmic scale concentration of Z-VAD-FMK at the micromolar scale was defined as the target. Three batches of independent experiments were performed, and two 384-well plates were used in each batch. Images at 0 - 12 hours after DMSO / Z-VAD-FMK treatment from two experimental batches (768 images for each class) were used for training and validation (training dataset), and images at specific time points after treatment from the remaining experimental batch were used for testing (128 images for each class; test dataset). The same cropping size as determined by the cell fate classification analysis was also adopted. For the evaluation of images of cells treated with other compounds and images of cells modified by CRISPR, all wells of the plate were used as the test dataset.

[0126] (Configuration of Convolutional Neural Network) The deep learning framework PyTorch (https: / / pytorch.org / ) (Paszke A. et al., Automatic differentiation in PyTorch. In NIPS (2017)) was used for network construction, training, validation, and testing. As the neural network architecture, the densenet-121 model was utilized (Huang G. et al., Proc. - 30th IEEE Conf. Comput. Vis. Pattern Recognition, CVPR 2017 2017-January, 2261-2269 (2016)). A fine-tuning approach effective in cases where the number of training samples is limited, as is often the case with medical or biological data, was adopted (Esteva A. et al., Nature 542, 115-118 (2017)). The network was pre-trained on the ImageNet dataset, which contains 1.2 million labeled images of objects in 1,000 categories. The number of nodes in the final layer of the pre-trained network was adjusted to the number of cell fate classes (= 2), or the number of nodes that output the predicted ZVAD-FMK-equivalent concentration or DL prediction score. The number of input nodes was adjusted according to the image size. During the training phase, the weights of all layers were re-optimized. The mean squared error (regression analysis) was used as the loss function. The error was backpropagated through the network, and the weights were optimized by stochastic gradient descent (SGD) using mini-batches.

[0127] (Training and Evaluation) First, the network was trained with fixed hyperparameters of SGD, learning rate lr = 0.001, momentum m = 0.9, evaluated by mean squared error, and compared among conditions of different cropping sizes of the input image (w x h = {256 × 256, 512 × 512, 768 × 768, 1,024 × 1,024}). After fixing the cropping size, grid search was performed for learning rate lr = {0.001, 0.002, 0.004} and momentum m = {0, 0.5, 0.9}, and the mean squared error was used to evaluate the training dataset by 4-fold cross-validation. Using the optimized hyperparameters, the network was trained and evaluated. The network parameters at the epoch when the network showed the highest classification accuracy for the training dataset were used for testing. The training and test experiment batches were rotated, and the average of the results of 3 trials was calculated. The training mini-batch size was b = 8, and the number of epochs during training was 10. The data was processed in parallel on a dual GPU. A straight line was fitted between the DL prediction score and the actual Z-VAD-FMK concentration, and the performance of the network was evaluated by the coefficient of determination (R-squared value).

[0128] (Visualization of Image Processing in CNN) To interpret where the focus of the CNN was when processing the input cell image, GradCAM was implemented in the CNN. GradCAM provides a gradient-based class activation map that enables the identification of the localized regions utilized by the CNN to make decisions (Selvaraju, R.R. et al., arXiv:1610.02391v1 (2016)). This method was applied to the trained CNN to obtain a map visualizing the focus of the CNN on the image when the network determined that the prediction score of the image was high. For more detailed interpretation, Guided GradCAM, which is a combination of the guided error backpropagation method and GradCAM, was also applied to the CNN (Springenberg et al., 2014).

[0129] (Human brain tissue) All postmortem human brain tissues from healthy controls and AD patients who had been pre-registered and consented prior to death were obtained from the Brain and Body Donation Program of the Banner Sun Health Research Institute (Beach T.G. et al., Cell Tissue Bank. 9, 229-245 (2008)), and the standard research protocol was approved by the Institutional Review Board. This study was approved by the Ethics Committee of the National Hospital Organization Tottori Medical Center. All controls and AD patients were diagnosed based on antemortem medical records and postmortem neuropathological examinations using the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) pathological criteria (Mirra, S.S. et al. Neurology 41, 479-486 (1991)) and the Braak staging (Braak H. et al., Acta Neuropathol. 112, 389-404 (2006)).

[0130] (Immunohistochemistry for human brain) For tissue staining, the brain was removed within 4 hours after death, coronal sections were prepared at 1 cm intervals, fixed with ice-cold 4% buffered paraformaldehyde (pH 7.4), cut into 40 μm sections, and stored at -30 °C in 2% DMSO / 20% glycerol with cryoprotection. Fixed floating tissue sections derived from the temporal cortex were treated under microwave irradiation in 0.1 M citrate buffer (pH 6.0), then treated with PBS containing 0.5% Triton X-100, 3% hydrogen peroxide, and 4% BSA at room temperature for 30 minutes, treated with a rabbit polyclonal antibody against KLHL32 (ab243805, Abcam, Cambridge, Massachusetts) diluted 1:200 at 4 °C for 24 hours, and then treated with biotin-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories, Bar Harbor, Maine). Subsequently, the labeled sections were reacted with an avidin-biotin-peroxidase complex (Vectastain Elite ABC kit, Vector Laboratories, Burlingame, California), and peroxidase activity was visualized with 0.05% 3,3'-diaminobenzidine tetrahydrochloride (Dojindo, Kumamoto, Japan) as previously reported (Kume H. et al., Neuropathol. Appl. Neurobiol. 35, 178-188 (2009); Tooyama I., et al., Dement. Geriatr. Cogn. Disord. 12, 237-242 (2001)). Sections stained with KLHL32 were digitized with a BIOREVO BZ-X800 microscope (Keyence, Osaka, Japan). The obtained images were quantified using imaging software cellSens (Olympus, Tokyo, Japan).

[0131] Example 1: Prediction of degenerating neurons in vitro using a deep learning-based iPSC model As shown in the figure, the objective of the present example was to develop a deep learning-based iPSC model (Deep-i model) that predicts the cell fate of iPSC-derived neurons using only phase difference images (Fig. 1A). Neuronal cell death is the most characteristic histopathological feature of neurodegenerative diseases including AD, and activated caspase-3 is a molecular marker of neuronal cell death in the brains of AD patients (Christie L.A. et al., Neurobiol. Dis. 26, 165-173 (2007); Louneva N. et al., Am. J. Pathol. 173, 1488-1495 (2008); Su J.H. et al., Brain Res. 898, 350-357 (2001)). It is widely known that neurons differentiated from iPS cells of patients with neurodegenerative diseases reproduce cell death over time (Imamura K. et al., Sci. Rep. 6, 34904 (2016); Imamura K. et al., Sci. Transl. Med. 9, eaaf3962 (2017); Nguyen H.N. et al., Cell Stem Cell 8, 267-280 (2011); Sanchez-Danes A. et al., EMBO Mol. Med. 4, 380-395 (2012)). Cortical neurons derived from iPSCs of familial AD also gradually show cell death in vitro (Kondo T. et al, Cell Stem Cell 12, 487-496 (2013)).

[0132] First, a live imaging system for evaluating neuronal cell death was constructed using iPSCs established from familial AD patients (Figure 1B). In iPSCs derived from familial AD patients, exogenous expression of human neurogenin 2 (NGN2), one of the specific transcription factors for cortical neuronal differentiation, was induced to induce differentiation into cortical neurons, and iPSC-derived cortical neurons were successfully prepared only 8 days after the start of differentiation (Figure 1B). The induction period of NGN2 expression was 5 days (120 hours). This differentiation method can convert human iPS cells into cortical neurons with almost 100% purity and high reproducibility (Kondo T. et al., Cell Rep. 21, 2304-2312 (2017)), and this high purity and reproducibility are necessary for analyzing and detecting subtle differences between different cell fates. To continuously evaluate the time-dependent change in caspase-3 activation, the 4-amino acid peptide DEVD conjugated with a nucleic acid-binding dye was used to detect activated caspase-3 as green fluorescence (Figure 1C). Eight days after the start of differentiation induction (192 hours after the start of NGN2 expression induction) was set as the starting point of analysis (i.e., the time point when the addition of DEVD was started) (Time 0). Time-lapse images taken every 6 hours showed that caspase-3 activation occurred over 7 days (Figures 1D and 1E). This result of cell death via activated caspase-3 was consistent with previous studies using iPSC-derived neurons from patients with neurodegenerative diseases (Imamura K. et al., Sci. Transl. Med. 9, eaaf3962 (2017); Kondo T. et al, Cell Stem Cell 12, 487-496 (2013)). Furthermore, to confirm that the green fluorescence in the experiment reflected caspase-3 activation, 25 μM of Z-Val-Ala-Asp(OMe)-CH2F (Z-VAD-FMK), a cell membrane-permeable inhibitor against caspases, was added. Almost complete disappearance of the green fluorescence was observed, which means that Z-VAD-FMK inhibited caspase-3 activation (Figures 1D and 1E). In the future, the difference in caspase-3 activation state between the conditions under which cells die and survive was statistically significant from 30 hours after the start of the assay (Figure 1E).

[0133] From these results, it was shown that the constructed live imaging system could evaluate the continuous increase in neuronal death caused by activated caspase-3 in a time-dependent manner. Using this live imaging system, a Deep-i model was established to predict the fate of neuronal death.

[0134] Next, Z-VAD-FMK was added to cortical neuron cultures at 8 concentrations (0, 0.0016, 0.008, 0.04, 0.2, 1, 5, 25 μM), and subsequent caspase-3 activation was monitored. Phase-contrast images and green fluorescence images were captured under different conditions every 6 hours (Figure 2A). Activated caspase-3 in neurons detected by enhanced green fluorescence decreased in a dose-dependent manner with Z-VAD-FMK (Figure 2A). Furthermore, it was also confirmed that there was no positional effect of the green fluorescence signal derived from activated caspase-3 among different columns of the 384-well plate (Figure 3A). To confirm the reproducibility of the experimental scheme, two additional batches showing the same results as the original batch 1 were prepared (Figures 4A and 4B). A dataset of phase-contrast images from a total of 3 batches was utilized as a cell death simulation system. To predict the fate of neurons before cell death became significantly apparent, the Deep-i model was trained (Figure 2B). In the training step, instead of a simple classifier for cell death or cell survival states, a deep learning-based prediction method was constructed as a way to calculate the probability of death as a continuous scale. As a neural network architecture, the DenseNet121 network was utilized (Huang G. et al., Proc. - 30th IEEE Conf. Comput. Vis. Pattern Recognition, CVPR 2017 2017-January, 2261-2269 (2016)), and the network was pre-trained on the ImageNet dataset (Deng J. et al., In 2009 IEEE Conference on Computer Vision and Pattern Recognition, (IEEE), pp. 248-255 (2009)). The network was re-trained to predict the concentration of Z-VAD-FMK, and the predicted concentration was defined as the DL prediction score. A dataset of phase-contrast images for each time frame from 0 to 72 hours after the start of tracking (Figure 5A) was prepared. Three batches of the experiment were conducted (Figure 5B), and two 384-well plates were used in each batch (Figure 5B).

[0135] For the training dataset, images were taken from two experimental batches between 0 and 12 hours after the start (Figure 4B). In total, 1,536 images within the time-window were prepared (Figure 4A). To learn from rich samples, all three different experimental batches, each consisting of two culture plates, were used, and the training, validation, and test steps were rotated among these independent batches to confirm reproducibility and avoid batch-dependent results (Figure 5B).

[0136] To evaluate the influence of different pixel sizes of the images, various pixel sizes of four square images were cropped from each full-size image (Figure 5C). The image size was optimized according to the mean squared error obtained from a four-fold cross-validation of the training dataset. In all three combinations of experimental batches (Figure 5B), the results of the largest pixel image, 1,024 × 1,024, had the smallest mean squared error between the actual Z-VAD-FMK concentration on a logarithmic scale and the predicted concentration defined as the DL prediction score (Figure 2C). Therefore, for further analysis, the cropping of 1,024 × 1,024 pixels was selected (Figure 4D).

[0137] After optimizing the input image size and hyperparameters, the performance of the trained network was evaluated against the test dataset. The network was trained using the majority of the training dataset, and the loss of the remaining training dataset was evaluated at each epoch. The best learning state of the network over the entire training epoch was saved. For testing, images from two independent batches used in the training step were used. The test dataset was images collected 24 hours after the start of tracking, and no significant difference in caspase-3 signal was observed between the DMSO and Z-VAD-FMK conditions (Figure 1E). The trained network generated DL prediction scores from the test images. The DL prediction scores showed a linear correlation with the actual Z-VAD-FMK concentration 24 hours after the start of the assay (Figure 2D).

[0138] The accuracy of the prediction at different time points when the training data was acquired was compared. The R-squared of the linear regression model for the scores was approximately 0.7 for all three batches (Figure 5D). When the test images were drawn from a later time point, the mean squared error increased (Figure 2E), and the R-squared of the approximate linear model decreased (Figure 2F). It was also confirmed that the DL prediction scores mainly originated from cell features and not from artifacts caused by the well positions. The DL prediction scores were inferred from images from plates where all wells at any position were in the DMSO state. The scores for all plates adjusted with DMSO did not show an obvious change according to the well positions (Figure 5E).

[0139] Next, the established system was applied to and verified for disease modeling. The PSEN1 G384A mutation in FAD patient iPSCs was genetically corrected, and isogenic iPSC clones were generated that retained the same genome as the original FAD iPSCs except for the sequence of genetically edited PSEN1 using the CRISPR-Cas9 system (Kondo T. et al., Cell Rep. 21, 2304-2312 (2017)). The trained CNN processed images from both FAD and isogenic iPSC-derived neurons and compared the DL prediction scores among different clones (Figs. 2G and 5F). Neurons without the PSEN1 mutation had higher DL prediction scores and a higher tendency to survive into the future. Furthermore, a compound that can modify Aβ production, which is one of the most important causes of AD and changes in FAD neurons, was tested. BSI IV, a 0-25 μM β-secretase inhibitor that suppresses Aβ production in a dose-dependent manner, was used. The DL prediction score increased in a dose-dependent manner with BSI IV, which showed a similar tendency to a mild suppression of cell death compared to the results of Z-VAD-FMK (Fig. 2H). These results indicated that the Deep-i model is applicable to disease modeling and that the DL prediction score is not unique only to the concentration of Z-VAD-FMK used for training the Deep-i model.

[0140] Furthermore, the applicability of the DL prediction score based on Z-VAD-FMK to other compounds was evaluated. First, necrostatin-1, a potent inhibitor of RIP1 kinase that suppresses necrotic cell death, was evaluated (Degterev A. et al., Nat. Chem. Biol. 1, 112-119 (2005)). As expected, compared with Z-VAD-FMK, necrostatin-1 partially rescued neurodegeneration and showed higher DL prediction scores in the medium to high concentration range (Figs. 2I and 2J). Also, a necrostatin-1 analog that can suppress neurodegeneration only under high-concentration conditions was evaluated (Takahashi N. et al., Cell Death Dis. 2012 311 3, e437-e437 (2012)), and it was found that the necrostatin-1 analog improved the DL prediction score only at high concentrations of 5-25 μM (Figs. 2K and 2L). Furthermore, DHA, which improves neurodegeneration via a non-caspase-inhibiting or non-RIPK pathway, was evaluated, and it was found that the changes in the DL prediction score at different concentrations showed a pattern similar to the changes in actual neurodegeneration shown as a fluorescence signal (Figs. 2M and 2N). The dose-dependence of the DL prediction score suggests that the score reflects the origin of cell degeneration that is independent of caspase 3. From these results, it was confirmed that the DL prediction score can reflect subtle changes in the neurodegenerative state due to different compounds and variable concentrations.

[0141] Example 2: Identification of molecular markers of cells programmed for neurodegeneration by single-cell RNA sequencing analysis As described above, the Deep-i model was able to detect hidden neurodegeneration during the process to neuronal death. However, while the Deep-i model can detect which cells begin to degenerate, it cannot detect the molecular signature of the cells. Therefore, to find the molecular signature of neurons in which neurodegeneration was programmed in vitro, single-cell RNA sequencing analysis was performed simultaneously. Twelve hours after the start of the assay (Time 0), it was still a stage where an increase in caspase-3 activation was not yet detected even in the negative control group without the addition of Z-VAD-FMK. At that stage, neurons were sampled and data from three different conditions, the DMSO control condition, the Z-VAD-FMK 1 μM condition, and the Z-VAD-FMK 25 μM condition, were compared (Figure 6A). After strict quality control filtering, 3,937 - 6,214 cells were obtained for each condition (see Figure 7A), and a median of 2,753, 2,484, and 2,202 genes were detected per cell for each condition (Figure 6B). It was determined that there was no obvious difference in the expression patterns of the overall genes or mitochondrial genes among the three different conditions and that there was no experimental noise that usually results from artificial cell damage (Figure 7A). To visualize the detailed heterogeneity, unsupervised clustering analysis was performed for each condition, and the single-nucleus transcriptome was classified into 15 clusters (UMAP plot, Figure 6B). As a result of comparing the percentage of cells in each cluster (%) among the three different conditions, it was revealed that the cluster size of "Cluster 12" (the proportion of cells included in Cluster 12 among the total number of cells) decreased in a dose-dependent manner with the increase in the concentration of Z-VAD-FMK (equivalent to the decrease in the number of apoptotic neurons) (Figure 6C). Cluster 12 contained 1.42% of the cells in the DMSO control group, 1.22% of the cells in the Z-VAD-FMK 1 μM treatment group, and 0.75% of the cells in the Z-VAD-FMK 25 μM treatment group, respectively (Figure 6C and 2B). This result indicates that the population of neurons in Cluster 12 is the cells that are destined to undergo neurodegeneration several hours later.

[0142] Next, to genetically characterize the cells contained in cluster 12, the gene set expressed by cluster 12 was analyzed. Molecular pathways associated with several genes were specific to cluster 12 (Figure 6D), and it was revealed that they could serve as markers for labeling cluster 12. In particular, KLHL32 was the most specific gene of cluster 12, and approximately 97% of the individual neurons in cluster 12 showed high expression of KLHL32 (Figure 6E). Furthermore, pathway analysis was performed on the most specific genes (top 50) of cluster 12, and the top two pathways associated with cluster 12 were found to be synapse formation and cell death (Figure 6D). To confirm the importance of KLHL32 in cell death events via activated caspase-3, siRNA knockdown experiments were conducted, and it was found that the suppression of KLHL32 inhibited the activation of caspase-3 and the fate of cell death (Figures 7C and 7D). The target sequences of the siRNAs used are shown in Table 1. Four types of siRNAs were mixed and used for each target gene (including Non-targeting Control). On the other hand, Z-Val-Ala-Asp(OMe)-CH2F (hereinafter, Z-VAD-FMK) also suppressed the expression of KLHL32 in a dose-dependent manner (Figures 6F and 6G). The cell bodies of KLHL32-positive cells were enlarged and showed a rounded shape, which resembled the shape of dying neurons. The positive rate of KLHL32 was approximately several percent (Figure 6G), and this positive rate was consistent with the abundance of cluster 12 in single-cell RNA sequencing analysis. From these results, KLHL32 was identified as an early marker before the fate of cell death at the single-cell level through analysis.

[0143]

Table 1

[0144] Furthermore, we combined the deep learning-based method with the results of single-cell RNA sequencing analysis to evaluate whether the Deep-i model could observe the KLHL32-positive population and predict the DL score. We applied the GradCAM (Gradient-weighted Class Activation Mapping) method (Selvaraju, R.R. et al., arXiv:1610.02391v1 (2016)), which can visually explain where the deep learning-based method observes and determines the DL score. We generated a GradCAM map for predicting the fate of neurodegeneration from Dense Block 3 in the DenseNet-121 architecture (Figure 8A). According to the original description (Selvaraju, R.R. et al., arXiv:1610.02391v1 (2016)), we binarized the map with a threshold of 15% of the maximum intensity and defined the area of the GradCAM-positive region (named "Grad") (Figures 8B and 3C). We also labeled the cell bodies of neurons (named "Cell") and the KLHL32-positive region (named "Mark"). By quantifying the overlapping regions between Grad, Cell, or Mark (Figures 8B and 8C), we were able to clarify that the Deep-i model examined the cell bodies of neurons rather than background noise (Figures 7C and 7D). As a result of examining the overlap with the GradCAM-positive region, more overlap with KLHL32-positive cells (Mark-Grad / Mark) was observed than with the entire cell body (Cell-Grad / Cell) (Figure 6H). These results indicated that the Deep-i model focuses on KLHL32-positive cells to predict the fate of neurons. From the above, the Deep-i model could explain what it observed as a molecular network and could identify KLHL32 as the molecular signature of cells in which neurodegeneration was programmed.

[0145] Example 3: Identification of preferential accumulation sites of cells programmed for neurodegeneration in a mouse model of Alzheimer's disease (AD) or postmortem brain tissues of AD patients The Deep-i model provides marker information to know when, where, and what is happening during the fate of neurodegeneration. Next, it was investigated whether the identified marker, KLHL32, is found during the process of neurodegeneration in the brains of AD mouse models or postmortem brain tissues of AD patients.

[0146] First, 5×FAD mice were used to investigate brain samples at a pre-stage before neurodegenerative death occurs (Eimer W.A. and Vassar R., Mol. Neurodegener. 8, 2 (2013); Oakley H. et al., J. Neurosci. 26, 10129-10140 (2006)). 5×FAD mice are one of the most widely used AD mouse models, and neurodegeneration is observed at about 4 to 6 months after birth (Eimer W.A. and Vassar R., Mol. Neurodegener. 8, 2 (2013)). Three months before neurodegeneration, the distribution of KLHL32-positive cells was observed. As previously reported, Aβ pathology was localized in the retrosplenial posterior cortex (RSP) and subiculum (SUB) (Figure 9A). The quantified regions of KLHL-32-positive cells were predominantly observed only in 5×FAD mice compared to control littermates (non-transgenic mice) (Figures 9B and 10). The distribution of KLHL32-positive cells was similar to Aβ pathology (Figures 9A and 9B). On the other hand, in the hippocampal CA1 (CA1) and entorhinal cortex (EN) where no Aβ pathology was observed at 3 months of age, KLHL32-positive cells were not observed except for astrocytes under the epithelium (Figures 9A and 9B). From these results, it was observed that KLHL32 is associated with the pathology of AD before the onset of neurodegeneration, and it became clear that it can be a marker for early identification of cells programmed for neurodegeneration that cause cell death and subsequent brain atrophy.

[0147] Next, it was examined whether KLHL32 is also distributed in the brains of AD patients. Postmortem brain samples from neuropathologically diagnosed AD and healthy elderly controls were used. When examining the immunopositivity of KLHL32, it was found that KLHL32 was positively stained in neurons and astrocytes in the temporal cortex (Figure 9C). Particularly in the case of the brains of AD patients, compared with healthy elderly controls (Figure 9E), KLHL32 was distributed in senile plaques, which are neuropathological features of AD, and in degenerated neurites indicating the pre-stage of neurodegeneration (Figure 9D). From these results, it is shown that the Deep-i model can identify markers for identifying cell populations destined for neurodegeneration as a result of AD pathology.

[0148] From the above results, it is strongly suggested that the drug that suppresses the expression of the KLHL32 gene exerts a preventive or therapeutic effect on neurodegeneration.

Industrial Applicability

[0149] The drug that suppresses the expression of the KLHL32 gene is useful for the prevention or treatment of neurodegenerative diseases. In addition, the drug that suppresses the expression of the KLHL32 gene is also useful as a combined drug for the treatment of neurodegenerative diseases.

Claims

1. A preventive or therapeutic agent for neurodegenerative diseases, comprising an agent for suppressing the expression of the KLHL32 gene.

2. The preventive or therapeutic agent according to claim 1, wherein the expression-suppressing agent is selected from the group consisting of siRNA, heteroduplex nucleic acid, antisense nucleic acid, shRNA, miRNA, antigene nucleic acid, and CRISPR-Cas system.

3. The preventive or therapeutic agent according to claim 2, wherein the expression-suppressing agent is siRNA.

4. The preventive or therapeutic agent according to any one of claims 1 to 3, wherein the neurodegenerative disease is selected from the group consisting of Alzheimer's disease, Parkinson's disease, amyotrophic lateral sclerosis (ALS), spinocerebellar degeneration, frontotemporal lobar degeneration, Lewy body dementia, multiple system atrophy, Huntington's disease, progressive supranuclear palsy, and corticobasal degeneration.

5. The preventive or therapeutic agent according to any one of claims 1 to 4, wherein the neurodegenerative disease is Alzheimer's disease.

6. (1) A step of contacting a test substance with cells expressing the KLHL32 gene, and (2) A step of selecting a test substance with a reduced expression level of the KLHL32 gene as a candidate for a preventive or therapeutic agent for neurodegenerative diseases A screening method for a preventive or therapeutic agent for neurodegenerative diseases, comprising:

7. (1) A step of contacting a test substance with cells, and (2) A step of selecting a test substance that suppresses an increase in the expression level of the KLHL32 gene as a candidate for a preventive or therapeutic agent for neurodegenerative diseases A screening method for a preventive or therapeutic agent for neurodegenerative diseases, comprising:

8. The method according to claim 6 or 7, wherein the cells are model cells for neurodegenerative diseases.

9. The method according to claim 8, wherein the model cells for neurodegenerative diseases are cells derived from a patient with a neurodegenerative disease or cells differentiated from pluripotent stem cells having a mutation in a gene that causes the neurodegenerative disease.

10. A biomarker for diagnosing neurodegenerative diseases, comprising the KLHL32 protein or KLHL32 transcript.

11. A method for assisting in the diagnosis of whether a subject has a neurodegenerative disease, comprising a step of detecting the biomarker according to claim 10 in a subject or a sample derived from the subject.

12. A kit for diagnosing neurodegenerative diseases, comprising an antibody that specifically recognizes the KLHL32 protein or a nucleic acid probe or nucleic acid primer that specifically recognizes the KLHL32 transcript.

Images