WO2009094592A2 - Genetic basis of alzheimer's disease and diagnosis and treatment thereof - Google Patents

Abstract

A collection of polymorphic sites conferring resistance or susceptibility to Alzheimer's disease and diseases related thereto is provided. The sites are useful in methods of diagnosing and treating Alzheimer's disease and related conditions.

Description

GENETIC BASIS OF ALZHEIMER'S DISEASE AND DIAGNOSIS AND TREATMENT THEREOF

STATEMENT AS TO RIGHTS TO INVENTIONS MADE UNDER FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

This invention was made with the support of the United States government under Contract numbers 1R44 AG024027, which was converted to ROl AG024027, by National Institute of Aging/National Institute of Health.

This application incorporates by reference herein in their entirety the following files submitted on duplicate compact disks:

File Name Date Created Size filtered_all_pO5_latest_v2.txt Jan. 23, 2008 233 KB filtered_40416_SNPs.txt Jan. 23, 2008 475 KB

EvsL_Logistic_l -5-07.txt Jan. 23, 2008 172 KB

AOO_lm_l-5-07.txt Jan. 23, 2008 243 KB

SNP information.txt Jan. 23, 2008 626 KB haplotype 3genes data tab.txt Jan. 23, 2008 116 KB data.txt Jan. 23, 2008 85 KB data detail-allele freq.txt Jan. 23, 2008 2 KB

The file contents of these files are as follows: filtered_all_pO5_latestjv2.txt:

This file contains a table, termed Table A in the specification of the instant application submitted herewith, that contains data identifying polymorphic loci associated with AD-related disease. filtered_40416_SNPs.txt:

This file contains a table, termed Table B in the specification of the instant application submitted herewith, that contains that contains data identifying polymorphic loci associated with AD-related disease. EvsL_Logistic_l-5-07.txt:

This file contains a table, termed Table C in the specification of the instant application submitted herewith, that contains data identifying polymorphic loci associated with age-of-onset of AD-related disease. AOO_lm_l-5-07.txt:

This file contains a table, termed Table D in the specification of the instant application submitted herewith, that contains data identifying polymorphic loci associated with age-of-onset of AD-related disease. SNP _information.txt:

This file contains a table, termed Table E in the specification of the instant application submitted herewith, that contains additional information about polymorphic loci associated with AD-related disease, including allele and sequence information. haplotype_3genes_data_tab.txt: This file contains a table that provides haplotype analyses for three genes: PSEN2,

APP, and HDAC4. data.txt:

This file contains a table that provides data underlying analyses for three genes: PSEN2, APP, and HDAC4. More information on these analyses are provided in Example 24. data_detail-allele_freq.txt:

This file contains a table that provides analyses for a set of haplotype alleles: gl-gl2. More information on these analyses are provided in Example 24.

THE MACHINE FORMAT FOR THE DUPLICATE COMPACT DISKS IS IBM-PC, AND THE OPERATING SYSTEM COMPATIBILITY IS MS-WINDOWS.

BACKGROUND OF THE INVENTION

Alzheimer's disease (AD) is a progressive disorder that gradually destroys a person's brain, including the brain's memory ability and ability to learn, reasoning, judgment, communication and ability to carry out daily activities. As AD progresses, individuals may also experience changes in personality and behavior, such as anxiety, depression, suspiciousness or agitation, infantile-like behavior, as well as delusions or hallucinations. The duration of the illness may vary from individual to individual. A person suffering from AD eventually requires complete care. If that individual does not die from other serious illness, complications from the AD and the loss of brain function itself can cause death. Drugs such as tacrine (Cognex), donepezil (Aricept), rivastigmine (Exelon), or galantamine (Reminyl) have been reported to help people in the early and middle stages of the disease or delay some symptoms. Another drug, memantine (Namenda), has been approved for treatment of moderate to severe AD. It has also been reported that non- inflammatory drags such as nonsteroidal anti-inflammatory drags (NSAIDs) help slow the progression of AD. Vitamin E has also been reported to slow the progress of AD.

BRIEF SUMMARY OF THE CLAIMED INVENTION The invention provides methods of polymorphic profiling. Such methods determine a polymorphic profile in an individual by determining the individual's genotype at one or more genetic loci associated with a phenotype of interest, e.g., susceptibility or resistance to an Alzheimer's disease-related disease (AD-related disease). Examples of such genetic loci are provided, e.g., in Tables A, B, C, D, or E. In certain preferred embodiments, a polymorphic profile is determined at one or more genetic loci within or proximal to at least one gene selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10orf35, LOC729099, NEUROG3, C10ORF35, LOC729099, ND3, and ND4, the latter two of which are encoded in the mitochondrial genome. Optionally, the group can further comprise other genes from Tables A, B, C and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, HDAC4, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, APP, or PSEN2. Optionally, the method comprises determining the total number of resistance and susceptibility alleles in a polymorphic profile, whereby the number of susceptibility alleles and/or the ratio of susceptibility alleles to resistance alleles provides an indication of whether the individual has or is at risk of developing an AD-related disease, or the likelihood of developing an AD-related disease at an early age. For example, a ratio of resistance to susceptibility alleles of less than a threshold value can be an indication that the individual is at high or low risk of developing Alzheimer's disease. A threshold value may be determined according to the methods provided in USSN 60/648,957, filed January 31, 2005; USSN 11/344,975, filed January 31, 2006, and PCT application no. US2006/003384, filed January 31, 2006. Optionally, a polymorphic profile is determined at polymorphic sites in or within 10 kb of at least ten genes selected from the group, and presence of at least twenty susceptibility and resistance alleles is determined. Optionally, a polymorphic profile is determined in an individual having a symptom of, or known susceptibility to, AD-related disease.

Optionally, a polymorphic profile can be determined in at least two but no more than 1000 different genomic regions (e.g., haplotype blocks or LD bins), at least two of the genomic regions including or overlapping a gene selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, APOC4, BCAM, LOC728050, NEUROG3, C10orf35, LOC729099, NEUROG3, C10ORF35, LOC729099, ND3, andND4, the latter two of which are encoded in the mitochondrial genome. Optionally, the group can further comprise other genes from Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4,

OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, AP0C4, APP, or PSEN2. Preferably, the at least two genomic regions in which the polymorphic profile is determined are at polymorphic sites in or within 10 kb of the at least two genes selected from the group. In some methods, the at least two genomic regions do not include APOE. In some methods, the at least two genomic regions each comprise or overlap at least one gene selected from the group consisting of LOC728050, NEUR0G3, C10orf35, LOC729099, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, and optionally, other genes from Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl , RAB 12, KIAA0802, CLYBL, ZIC5,

LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, AP0C4, APP, or PSEN2. In some methods, the polymorphic profile is determined in at least ten genomic regions, each including a different gene selected from the group. In some methods, the polymorphic profile is determined in at least two and no more than 50 different genomic regions. Some methods also involve selecting a treatment or prophylactic regime for an AD-related disease based on the polymorphic profile.

The invention further provides methods of diagnosing or prognosticating AD-related disease in a subject. Such methods comprise determining a polymorphic profile of a subject at genetic loci within or proximal to a gene selected from the group consisting of APOE,

APOCl, PVRL2, TOMM40, CLPTMl, APOC2, AP0C4, BCAM, LOC728050, NEUR0G3, C10orf35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes from Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB 12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, APP, or PSEN2. The AD-related disease can be, for example, early-onset AD, late-onset AD, or familial AD (FAD), and the presence of a susceptibility allele shown in Table C or in linkage disequilibrium therewith is an indication of a presence or susceptibility to AD-related disease, or the likelihood of developing an AD-related disease at an early age in the subject

The invention further provides methods of diagnosing or prognosticating AD-related disease that comprise determining a polymorphic profile in a genomic region (e g , haplotype block or LD bm) that includes or overlaps a gene selected from the group consisting of 5 APOE, APOCl, PVRL2, TOMM40, CLPTMl , APOC2, APOC4, BCAM, LOC728050, NEUROG3, C10orf35, LOC729099, ND3, and ND4 Optionally, the group can further comprise other genes from Tables A, B, C, and/or D, e g , PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EX0C2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932 Some methods

10 determine presence of a susceptibility allele shown in Table C or in linkage disequilibrium therewith, the susceptibility allele indicating presence or susceptibility to the AD-related disease, e g , late-onset Alzheimer's disease, or the likelihood of developing an AD-related disease at an early age

The invention further provides methods of diagnosing or prognosticating an AD-

15 related disease in a patient Such methods determine presence of at least one susceptibility allele shown m Table C or m linkage disequilibrium therewith, the presence of the susceptibility allele indicating presence or susceptibility to the AD-related disease Optionally, the method determines presence of at least one susceptibility allele shown in Table C Optionally, the method determines at least one susceptibility allele not in or within

20 40 kb of the APOE gene The AD-related disease may be, e g , early-onset, familial, or late- onset Alzheimer's disease

Any of the above methods can include informing the patient or a relative thereof of presence or susceptibility to an AD-related disease, performing a secondary test for an AD- related disease, such as determimng mental activity by a psychometric measure or taking a

25 biopsy, and/or administering a regime effective to treat or effect prophylaxis of an AD-related disease Any of the above methods can also involve determining at least one susceptibility allele not in or within 40 kb of TOMM40 or APOCl, or not in or within 40 kb of PVRL2, TOMM40, APOCl, APOC4, BCAM, APOC2, or CLPTMl

Optionally, any of the above methods determine presence of at least 5 or 10

30 susceptibility alleles at genetic loci within or proximal to at least five different genes selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUROG3, C10ORF35, LOC729099, ND3, and ND4, preferably susceptibility alleles are alleles shown Table C Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e g , PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, APP, or PSEN2. The invention further provides methods of expression profiling. Such methods entail determining expression levels of at least 2 and no more than 10,000 genes in a subject, wherein at least two of the genes are selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, the expression levels forming an expression profile. Optionally, the group can further comprise other genes in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, APP, or PSEN2. Optionally, the methods determine expression levels of the genes in a control subject free of an AD-related disease. Optionally the methods determine expression levels of the genes in a control subject having an AD-related disease. Optionally, the methods compare the expression levels of the genes in the subject with expression levels of the genes in a control subject known to have an AD-related disease and/or a control subject known to lack an AD-related disease, wherein similarity of expression profiles in the subject and the control subject having the AD-related disease is an indication the subject has the AD-related disease, or the likelihood of developing an AD- related disease at an early age. Likewise, similarity of the expression profiles in the subject and the control subject not having the AD-related disease is an indication the subject lacks presence of or susceptibility to the AD-related disease, or the likelihood of developing an AD-related disease at an early age.

The invention further provides a transgenic non-human animal comprising a genome comprising a transgene comprising an exogenous nucleic acid encoding the protein of a gene selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, AP0C4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, whereby the animal expresses the gene, and is disposed to develop at least one sign or symptom of an AD-related disease or an early age-of-onset thereof. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932.

The invention further provides a transgenic non-human animal comprising a genome comprising a transgene comprising an exogenous nucleic acid encoding the protein encoded by a gene selected from the group provided in Tables A and B, whereby the animal expresses the gene, and is disposed to develop at least one sign or symptom of an AD-related disease or an early age-of-onset thereof. In certain embodiments, the gene is selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORP35, LOC729099, ND3, andND4, or is a gene in linkage disequilibrium therewith. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the susceptibility allele is shown in Table C.

The invention further provides a transgenic non-human animal comprising a genome having an enhanced, inhibited, or disrupted endogenous gene that is the cognate form of a human gene provided in Tables A, B, C, and/or D, e.g., selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, whereby the transgenic- nonhuman animal develops at least one sign or symptom of an AD-related disease or an early age-of-onset thereof. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932.

The invention further provides a method for producing a transgenic knock-out or knock-in non-human animal. The method entails providing a targeting construct containing a disrupted segment of a gene provided in Tables A, B, C, and/or D, e.g., selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, and homologously recombining the targeting construct with the genome of a cell of the animal, whereby the construct is stably integrated into the genome of the cell; and propagating a transgenic animal from the cell. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. The invention further provides a method for producing a transgenic non-human animal. The method entails introducing a construct encoding and capable of expressing the protein encoded by a gene provided in Tables A, B, C, and/or D, e.g., selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, into a cell, and propagating a transgenic animal from the cell. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. The invention further provides a method for identifying an agent for use in diagnosis, prognosis, prophylaxis, or treatment, of an AD-related disease. The method entails contacting a polypeptide (or polypeptide fragment) encoded by a gene provided in Tables A, B, C, and/or D, e.g., selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, or a nucleic acid encoding the polypeptide, with an agent to be tested; assessing a level of binding of the agent to the polypeptide or a level of modulation of activity or expression of the polypeptide by the agent; and comparing the level of binding activity or expression of the polypeptide with a control sample in an absence of the agent, wherein a difference in level of binding, activity or expression in the presence of the agent relative to the control sample is an indication that the agent has activity useful in diagnosis, prognosis, prophylaxis, or treatment, an AD-related disease. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the polypeptide is an isolated polypeptide. Optionally, the polypeptide is expressed in a cell transformed with a nucleic acid encoding the polypeptide. Optionally, the method also involves determining whether the agent shows activity inhibiting development of or clearing a sign or symptom of the AD-related disease in an animal model. Optionally the assessing involves contacting the agent with the polypeptide and detecting specific binding between the compound and the polypeptide or detecting a modulation of activity of the polypeptide or detecting a modulation of expression of the polypeptide. Optionally, the assessing involved performing a clinical trial.

The invention further provides methods of effecting treatment or prophylaxis of an AD-related disease. Such methods comprise administering to the subject an effective amount of an agent that modulates the activity or expression of a protein encoded by a gene provided in Tables A, B, C, and/or D, e.g., selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, APP, or PSEN2. Optionally, the agent is selected from the group consisting of: an antibody, small molecule or natural product that specifically binds to a protein encoded by a gene selected from the group; a zinc finger protein that modulates expression of a gene selected from the group; or an siRNA, antisense RNA, RNA complementary to a regulatory sequence, or ribozyme that inhibits expression of a gene selected from the group. Optionally, the method also involves monitoring a sign or symptom of the AD-related disease in the patient responsive to the administration. Optionally, the method involves administering a second agent effective to effect treatment or prophylaxis (which includes delaying age-of-onset) of the AD-related disease. In some methods, the patient is human. In some methods, the disease is late-onset Alzheimer's disease.

The invention further provides a computer-implemented method of identifying a polymorphic profile characterizing a patient as amenable to treatment with an agent. Some methods involve providing data for a first population of patients with an AD-related disease treated with the agent and a second population of patients with the AD-related disease treated with a placebo, the data comprising whether the patient reached a desired endpoint, and a polymorphic profile of the patients in the first and second populations in at least one polymorphic site at genetic loci within or proximal to a gene provided in Tables A, B, C, and/or D, e.g., selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, and selecting first and second subpopulations from the first and second populations based on similarity of the polymorphic profile; and comparing the percentage of patients in the first subpopulation reaching the desired endpoint with the percentage of patients in the second subpopulation, a significant different indicating that the polymorphic profile of the subpopulations characterizes a patient as amenable to treatment. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, APP, or PSEN2.

The invention further provides a method of screening an agent for activity in treating an AD-related disease comprising performing a primary screen to determine whether the agent affects a level of expression or function of a protein encoded by a gene provided in Tables A, B, C, and/or D, e.g., selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, and performing a secondary screen to determine whether the agent affects (e.g., delays or prevents onset of) the AD-related disease in an animal.

Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl , AP0C2, APOC4, APP, or PSEN2. Optionally, the primary screen measures binding of the agent to the protein. Optionally, the primary screen measures capacity of the agent to agonize or antagonize the protein.

The invention further provides a method of screening an agent for activity in treating an AD-related disease comprising exposing a transgenic animal, e.g., such as those described herein, to the agent and determining whether the agent treats or inhibits further development of the disease, or delays or prevents onset of the disease, in the animal model.

The invention further provides a method for identifying a polymorphic site correlated with Alzheimer's disease or susceptibility thereto or age-of-onset thereof, comprising identifying a polymorphic site within a protein encoded by a gene provided in Tables A, B, C, and/or D, e.g., selected from the group consisting APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, AP0C4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4, and determining whether a variant polymorphic form occupying the site is associated with the disease or susceptibility thereto. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932.

The invention further provides a method of excluding an individual from a clinical trial to test a drug for treatment or prophylaxis of Alzheimer's disease. Such a method entails determining a polymorphic profile in an individual presenting symptoms resembling Alzheimer's disease in or within 10 kb of a plurality of genes selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4; determining the total number of resistance and susceptibility alleles at each locus in the polymorphic profile, wherein a high ratio of resistance to susceptibility alleles is an indication the individual should be excluded from the clinical trial. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, AP0C4, APP, or PSEN2.

The invention further provides for use of (a) an isolated nucleic acid that specifically hybridizes to a segment in the human genome that includes a single nucleotide polymorphism (SNP) at a position shown in column 9 of Table C or is in linkage disequilibrium therewith, (b) a SNP shown in column 9 of Table C or in linkage disequilibrium therewith, or (c) a protein encoded by the nucleic acid, or (d) an antibody that specifically binds to the protein for diagnosis, prognosis, prophylaxis, treatment or study of an AD-related disease. In some uses, the segment is located no further than 10 kb from the SNP. In some uses, the segment is within a gene including the SNP or in linkage disequilibrium therewith. Optionally, the isolated nucleic acid is a probe or primer. Optionally, the isolated nucleic acid is a cDNA. Optionally, the isolated nucleic acid is a gene shown in Tables A, B, C, and/or D. Optionally, the segment is not within the human ApoEl gene or in linkage disequilibrium therewith. Optionally, the disease is late-onset Alzheimer's disease. Optionally, the segment is within a gene selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EX0C2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, APOC4, APP, or PSEN2.

The invention further provides an isolated protein encoded by a gene shown in Tables A, B, C, and/or D. Optionally, at least one amino acid of the gene is encoded by a codon that includes a variant form of a polymorphic site shown in column 4 or 5 of Table C. Optionally, the gene is selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, APP, or PSEN2.

The invention further provides an antibody that specifically binds to a protein encoded by a gene selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes shown in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB 12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, AP0C4, APP, or PSEN2. Optionally, an amino acid of the protein is encoded by a nucleic acid in which an SNP shown in Table C is occupied by the nucleotide of Allele 1 and not the nucleotide of Allele 2, vice versa.

DEFINITIONS The term "a" or "an" as used herein may mean one or more. The term "Alzheimer's Disease" or "AD" is defined broadly to include asymptomatic as well as symptomatic conditions of Alzheimer's disease including genetic predisposition for AD, environmentally induced AD, early-onset AD, late-onset AD (LOAD), familial AD (FAD).

The term "AD-related disease" refers to one or more diseases, conditions or symptoms or susceptibility to diseases, conditions or symptoms that involve directly or indirectly, neurodegeneration including but not limited to the following: Alzheimer's disease (AD), amyotrophic lateral sclerosis (ALS), Alpers' disease, Batten disease, Cockayne syndrome, corticobasal ganglionic degeneration, Huntington's disease, Lewy body disease, Pick's disease, motor neuron disease, multiple system atrophy, olivopontocerebellar atrophy, Parkinson's disease, postpoliomyelitis syndrome, prion diseases, progressive supranuclear palsy, Rett syndrome, Shy-Drager syndrome and tuberous sclerosis, and may be characterized by, e.g., dementia, memory loss, confusion, and other neurodegenerative conditions. In certain aspects, an AD-related disease is a neurodegenerative disease the affects neurons in the brain. An AD-related disease may be e.g. a condition that is a risk factor for developing AD, or may be a condition for which AD is a risk factor, or both. Many AD-related diseases are characterized by related pathology of amyloid deposits of a protein that stain with Congo red dye and/or related neurodegenerative symptoms.

Susceptibility to AD or a related disease means that a subject has a significantly greater risk of developing the disease than the average risk of an age-, and sex-matched individual from the general population. Susceptibility can also mean that a subject is likely to exhibit a significantly earlier age-of-onset than the average age-of-onset of an age-, and sex- matched individual from the general population who will eventually develop the disease.

Resistance to AD or related disease means that a subject has a significantly lower risk of developing the disease than the average risk of an age- and sex-matched individual from the general population. Resistance can also mean that a subject is likely to exhibit a significantly later age-of-onset than the average age-of-onset of an age-, and sex-matched individual from the general population who will eventually develop the disease.

A nucleic acid or polypeptide associated with susceptibility (e.g., a susceptibility allele) to a disease is a nucleic acid or polypeptide that occurs significantly more frequently in a population of individuals having the disease, or having an earlier age-of-onset of the disease, than in a population of individuals lacking the disease.

A nucleic acid or polypeptide associated with resistance to a disease (e.g., a resistance allele) is a nucleic acid or polypeptide that occurs significantly less frequently in a population of individual having the disease, or having an earlier age-of-onset of the disease, than in a population of individuals lacking the disease.

A symptom of a disorder means a phenomenon experienced by an individual having the disorder indicating a departure from normal function, sensation, or appearance.

A sign of a disorder is any bodily manifestation that serves to indicate presence or risk of a disorder.

The term "AD nucleic acid" or "AD-associated genomic region" means a nucleic acid, or fragment, derivative, variant, polymorphism, or complement thereof, associated with resistance or susceptibility to AD-related disease or age-of-onset thereof, including, for example, at least one or more AD polymorphisms, genomic regions spanning 10 kb immediately upstream and 10 kb immediately downstream of an AD polymorphism, coding and non-coding regions of an associated gene, and/or genomic regions spanning 10 kb immediately upstream and 10 kb immediately downstream of an associated gene, and variants thereof. The term also includes nucleic acids similarly related to genes in an associated gene pathway. An AD nucleic acid may also be an "associated genomic region" when it is found within the genome of an organism.

The term "AD polymorphism" or "associated polymorphism" refers to a specific nucleic acid locus at which a nucleotide polymorphism associated with AD-related disease occurs. For example, an AD polymorphism may be a SNP position such as those provided in Table C. There may be two or more nucleotide base variants ("alleles") at a given AD polymorphism, and each of these alleles may be specifically associated with either a resistance or a susceptibility to AD-related disease, or to a response to a treatment regimen (e.g., drug response). An allele that is the same as that found in a reference nucleic acid sequence is referred to as a "reference allele," and an allele that is different than that found in the reference sequence is referred to as an "alternate allele."

The term "AD polypeptide" or "associated polypeptide" refers to any peptide, polypeptide, or fragment, derivative or variant thereof, associated with resistance or susceptibility to AD-related disease, including a peptide or polypeptide regulated or encoded, in whole or in part, by an associated gene or genomic regions of 10 kb immediately upstream and downstream of an associated gene, or fragment, variants, derivative, or modifications thereof. The term also includes such polypeptides up- or down-stream in an associated gene pathway.

The term "another" as used herein may mean at least a second or more.

The term "associated gene" or "associated gene region" or "AD gene" refers to a gene, a genomic region 10 kb upstream and 10 kb downstream of such gene, or regulatory regions that modulate the expression of such gene, comprising at least a portion of one of the polymorphic regions identified in Tables A, B, C, D, and/or E, and all associated gene products (e.g., isoforms, splicing variants, and/or modifications, derivatives, etc.) The sequence of an AD gene in an individual may contain one or more reference or alternate alleles, may contain a combination of reference and alternate alleles, or may contain alleles in linkage disequilibrium with one or more of the polymorphic regions identified in Tables A, B, C, D, and/or E.

The term "associated gene pathway" generally refers to genes and gene products comprising an AD-related disease pathway (i.e., a pathway related to resistance or susceptibility to AD-related disease), and may include one or more genes that act upstream or downstream of an associated gene in an AD-related disease pathway; or any gene whose product interacts with, binds to, competes with, induces, enhances or inhibits, directly or indirectly, the expression or activity of an associated gene; or any gene whose expression or activity is induced, enhanced or inhibited, directly or indirectly, by an associated gene. An associated gene pathway may refer to one or more genes.

The term "complementary" can mean partially complementary or completely complementary and generally refers to the natural hydrogen bonding between purine and pyrimidine base pairs. The term "partially complementary" refers to instances where only some of the base pairs are bonded. The term "completely complementary" refers to instances where all or nearly all of the base pairs are bonded. The term "perfectly complementary" refers to instances where all of the base pairs are bonded.

The term "derivative" refers to chemical modification of a nucleic acid, a protein or mimetic thereof. Examples of chemical modifications of a nucleic acid include replacement of hydrogen by an alkyl, an acyl or an amino group. A nucleic acid derivative may also refer to a nucleic acid that was derived from another nucleic acid (e.g., mRNA transcribed from a gene, cDNA synthesized from an RNA molecule, or cRNA synthesized from a DNA molecule, etc.) A nucleic acid derivative can encode a polypeptide that retains, changes, inhibits or enhances essential characteristics or functions of the polypeptide that the natural nucleic acid encodes. A polypeptide derivative is one that is modified by glycosylation, pegylation or other process, and that retains, changes, inhibits or enhances at least one characteristic or function (e.g., immunological response) of the polypeptide from which it was derived.

The term "stringent conditions" refers to conditions for hybridization of complementary nucleic acids wherein the presence of a nucleic acid may be detected. For example, the detection of hybridization may be used as a proxy for determining the presence of a particular nucleic acid. Different stringency conditions may be utilized under different circumstances. Stringent conditions depend on, for example, length of the nucleic acids, hybridization temperature, buffers, and other hybridization reaction conditions. Generally, stringent conditions are selected to be about 5°C lower than the thermal melting point (Tm) of a specific sequence at a defined ionic strength and pH. The Tm is the temperature (under defined ionic strength, pH and nucleic acid concentration) at which 50% of the complementary nucleic acids hybridize to a target nucleic acid at equilibrium. As target nucleic acids are generally present in excess, at Tm, 50% of the complementary nucleic acids are occupied at equilibrium. Typically, stringent conditions include a salt concentration of at least about 0.01 to 1.0 M Na ion concentration (or other salts) at pH 7.0 to 8.3 and the temperature is at least about 30°C for short probes (e.g., 10 to 50 nucleotides). Stringent conditions can also be achieved with the addition of destabilizing agents such as formamide. For example, in some embodiments, conditions of 5X SSPE (750 mM NaCl, 50 mM NaPhosphate, 5 mM EDTA, pH 7.4) and a temperature of 25-3O⁰C are suitable for allele- specific nucleic acid hybridizations. In other embodiments, conditions of IM TMACl (tetramethylammonium chloride), 3.25 M Tris (pH 7.8-8.0), 0.00325% Triton X-100, and a temperature of 50⁰C are suitable for allele-specific nucleic acid hybridizations. Example 16 provides yet another example of conditions appropriate for allele-specific nucleic acid hybridizations.

The terms "isolated" and "purified" refer to a material that is substantially or essentially removed from or concentrated in its natural environment. For example, an isolated nucleic acid is one that is separated from the nucleic acids that normally flank it or from other biological materials (e.g., other nucleic acids, proteins, lipids, cellular components, etc.) in a sample. In another example, a polypeptide is purified if it is substantially removed from or concentrated in its natural environment.

The term "nucleic acid," refers to a deoxyribonucleotide or ribonucleotide, whether singular or in polymers, naturally occurring or non-naturally occurring, double-stranded or single-stranded, translated (e.g., gene) or untranslated (e.g. regulatory region), or any fragments, derivatives, mimetics or complements thereof. A nucleic acid includes analogs (e.g., phosphorothioates, phosphoramidates, methyl phosphonate, chiral-methyl phosphonates, 2-O-methyl ribonucleotides) or modified nucleic acids (e.g., modified backbone residues or linkages) or nucleic acids that are combined with carbohydrates, lipids, protein or other materials, or peptide nucleic acids (PNAs) (e.g., chromatin, ribosomes, transcriptosomes, etc.) A nucleic acid can include one or more polymorphisms, variations or mutations (e.g., SNPs, insertions, deletions, inversions, translocations, etc.) Examples of nucleic acids include oligonucleotides, nucleotides, polynucleotides, nucleic acid sequences, genomic sequences, antisense nucleic acids, DNA regions, probes, primers, genes, regulatory regions, introns, exons, open-reading frames, binding agents, target nucleic acids and allele- specific nucleic acids.

A polymorphic or variant site is a locus of genetic variation in a genome or a location of amino acid variation in a protein. A polymorphic site is occupied by two or more polymorphic forms (also known as variant forms or alleles). A single nucleotide polymorphic site (SNP) is a variation at a single nucleotide. The term "polymorphism" refers a position in a nucleic acid or polypeptide that possesses the quality or character of occurring in several different forms. A nucleic acid or polypeptide may be naturally or non-naturally polymorphic, e.g., having one or more sequence differences (e.g., additions, deletions and/or substitutions) as compared to a reference sequence. A reference sequence may be based on publicly available information (e.g., the U.C. Santa Cruz Human Genome Browser Gateway (genome.ucsc.edu/cgi-bin/hgGateway) or the NCBI website (www.ncbi.nlm.nih.gov)) or may be determined by a practitioner of the present invention using methods well known in the art (e.g., by sequencing a reference nucleic acid). A nucleic acid polymorphism is characterized by two or more "alleles", or versions of the nucleic acid sequence. Typically, an allele of a polymorphism that is identical to a reference sequence is referred to as a "reference allele" and an allele of a polymorphism that is different from a reference sequence is referred to as an "alternate allele", or sometimes a "variant allele". However, as any two reference sequences may differ at a polymorphic locus, an "alternate allele" may be found in a reference sequence and a "reference allele" may not. Furthermore, the designation of a "reference allele" and an "alternate allele" need not be based on any particular reference different alleles of a polymorphism. As such, the designation of alleles provided herein as "reference" or "alternate" should not be construed to indicate that the allele is or is not present in a particular reference sequence. A nucleic acid comprising an alternate allele may be referred to as a "variant nucleic acid". Nucleic acid polymorphisms include loci within nucleic acids encoding a polypeptide, but which due to the degeneracy of the genetic code are not found in nature. A polypeptide polymorphism is characterized by two or more versions of an amino acid sequence, with a version that is identical to a reference sequence referred to as a "reference polypeptide" and a version that is different from a reference sequence referred to as an "alternate polypeptide" or a "polypeptide variant". Polypeptide polymorphisms include polypeptides encoded by another locus in the human genome or other organism's genome that have substantial homology, in whole or in part, to the polypeptides provided herein. The term "synonymous polymorphism" refers to a polymorphism in a coding region of a gene for which different alleles of the polymorphism encode an identical amino acid sequence. The term "non-synonymous polymorphism" refers to a polymorphism in a coding region of a gene for which different alleles of the polymorphism encode different amino acid sequences. Non-synonymous polymorphisms may be conservative or non-conservative. A "conservative polymorphism" refers to a non-synonymous polymorphism for which the different amino acid sequences encoded are functionally equivalent. A "non-conservative polymorphism" refers to a non-synonymous polymorphism for which the different amino acid sequences encoded are functionally dissimilar. "Functionally equivalent" as used herein refers to a polypeptide capable of exhibiting a substantially similar activity as another polypeptide.

The terms "polypeptide," "peptide," "oligopeptide" and "protein" are used interchangeably to refer to a polymer of amino acids, PNAs or mimetics, of no specific length and to all fragments, isoforms, variants, derivatives and modifications thereof A polypeptide may be naturally and non-naturally occurring The term variant when used to describe a polypeptide refers to variations in amino acid sequences as compared to a reference polypeptide sequence, whether or not such variations are encoded by conservative or non- conservative polymorphisms, for example An ammo acid substitution that is encoded by a conservative polymorphism may be referred to as a conservative substitution Likewise, an ammo acid substitution that is encoded by a non-conservative polymorphism may be referred to as a non-conservative substitution The term modification include tags, labels, post- translational modifications or other chemical or biological modifications In preferred embodiment a polypeptide is purified

The term "probes" or "primers" refers to nucleic acids that can hybridize, m whole or in part, in a base-specific manner to a complementary strand Typically, the term "primer" refers to a smgle-stranded nucleic acid that acts as a point of initiation of template-directed DNA synthesis (e g , PCR primers) and the term "probe" refers to a single-stranded nucleic acid designed to hybπdize to a target nucleic acid For example, hybridization of the probe to a sample nucleic acid may be used to purify a target nucleic acid within the sample, or detection of hybridization (or lack thereof) of the probe to a sample nucleic acid may be used to determine the presence (or absence) of a target nucleic acid in the sample Although smgle-stranded probes and pπmers are primarily discussed herein, the present invention is not limited to such probes and primers, double-stranded or partially double-stranded probes or primers are also included

The term "specific hybridization" refers to the ability of a first nucleic acid to bind, duplex or hybndize to a second nucleic acid in a manner such that the second nucleic acid can be identified or distinguished from other components of a mixture (e g , cellular extracts, genomic DNA, etc ) In certain embodiments, specific hybridization is performed under stringent conditions

The term "substrate" refers to any rigid or semi-rigid support to which molecules (e g , nucleic acids, polypeptides, mimetics) may be bound Examples of substrates include membranes, filters, chips, slides, wafers, fibers (e g , optical fibers), magnetic or nonmagnetic beads, gels, capillaries, or other tubmg, plates, polymers, and microparticles with a variety of surface forms including wells, trenches, pins, channels and pores, and may be manufactured from various substances, including but not limited to glass, silicon, fused silica, borosihcate, quartz, soda lime glass, a polymeric material (e g , polyethylene, polycarbonate, polyvmylchloπde, polystyrene, and the like) or a combination thereof The term "vector" refers to any construct or composition by which the expression, transfer or manipulation of a nucleic acid may be accomplished or facilitated. For example, the term vector can be an artificial chromosome (e.g., BAC, YAC, etc.), cosmid, viral particle, viral nucleic acid, plasmid, or a liposome. For example, in some embodiments a vector is a viral nucleic acid or a plasmid with appropriate transcription/translation control signals. An expression vector is a vector that is designed to promote the expression of one or more nucleic acid inserts.

The term "haplotype block" (as used herein, also referred to as a linkage disequilibrium bin) refers to a region of a chromosome that contains one or more polymorphic sites (e.g., 1-10) that tend to be inherited together (i.e., are in linkage disequilibrium) (see Patil, et al., Science, 294:1719-1723 (2001); US 20030186244)). In other words, combinations of local polymorphic forms that are correlated more often than expected by chance, sometimes at least greater than about 80% of the time in a population of individuals, and/or a set of polymorphic forms that together comprise an associated polymorphism. See, e.g., C. S. Carlson et al., Am. J. Hum. Genet. 74, 106 (2004); and Hinds, et al., Science 307, 1072 (2005). For example, combinations of polymorphic forms at the polymorphic sites within a block or linkage disequilibrium bin cosegregate in a population more frequently than combinations of polymorphic sites that occur in different haplotype blocks. The term "haplotype pattern" refers to a combination of polymorphic forms that occupy polymorphic sites, usually SNPs, in a haplotype block or linkage disequilibrium bin on a single DNA strand. For example, the combination of variant forms that occupy all the polymorphisms within a particular haplotype block on a single strand of nucleic acid is collectively referred to as a haplotype pattern of that particular haplotype block. Many haplotype blocks are characterized by four or fewer haplotype patterns in at least 80% of individuals. The identity of a haplotype pattern can often be determined from one or more haplotype determining polymorphic sites without analyzing all polymorphic sites constituting the pattern.

The term "linkage disequilibrium" refers to the preferential segregation of a particular polymorphic form with another polymorphic form at a different chromosomal location more frequently than expected by chance. Linkage disequilibrium can also refer to a situation in which a phenotypic trait displays preferential segregation with a particular polymorphic form or another phenotypic trait more frequently than expected by chance. The boundaries of a gene are defined by the beginning and end of its transcribed region. A polymorphic site is proximal to a gene if it occurs within the intergenic region between the transcribed region of the gene and that of an adjacent gene. Usually, proximal implies that the polymorphic site occurs closer to the transcribed region of the particular gene that that of an adjacent gene. Typically, proximal implies that a polymorphic site is within 400 kb, and preferably within 10 kb of the transcribed region. Polymorphic sites not occurring in proximal regions as defined above are said to occur in regions that are distal to the gene. If a segment of genomic DNA is said to occur within a certain distance of a polymorphic site, then the most distant point of the segment occurs within the specified distance. Likewise if a segment of genomic DNA is said to occur within a certain distance of a gene, then the most distant point of the segment occurs with the specified distance of closest transcriptional endpoint of the gene.

The term "specific binding" refers to the ability of a first molecule (e.g., an antibody) to bind or duplex to a second molecule (e.g., a polypeptide) in a manner such that the second molecule can be identified or distinguished from other components of a mixture (e.g., cellular extracts, total cellular polypeptides, etc.)

A nonhuman homolog (or cognate form) of a human gene is the gene in a nonhuman species, such as a mouse, that shows greatest sequence identity at the nucleic acid and encoded protein level, and higher order structure and function of the protein product to that of the human gene or encoded product. The terms "modulate" and "modulation" refer to a change such as in expression, lifespan, or function such as an increase, decrease, alteration, enhancement or inhibition of expression or activity of a gene or gene product.

Statistically significant" means significant at a p value < 0.05.

The term "comprising" indicates that other elements can be present besides those explicitly stated.

Various embodiments and modifications can be made to the invention disclosed in this application without departing from the scope and spirit of the invention. Unless otherwise apparent from the context any embodiment, feature or element of the invention can be used in combination with any other. Any embodiment, feature or element of the invention described in the alternative to other embodiments, features or elements can be excluded from the invention. Throughout this disclosure various patents, patent applications, gene identifiers, and publications are referenced and unless otherwise indicated, are incorporated by reference in their entirety and for all purposes to the same extent as if so individually denoted. DETAILED DESCRIPTION OF THE INVENTION

The invention provides a collection of polymorphic sites having resistance and susceptibility alleles associated with resistance or susceptibility (including age-of-onset) to Alzheimer's disease and other AD-related diseases, particularly the most common form of Alzheimer's disease, known as late-onset Alzheimer's disease (LOAD). The polymorphic sites were identified by analyzing a sampling of polymorphic sites throughout the human genome in a population having late-onset disease and a control population. This application hereby incorporates the following applications by reference in their entireties for all purposes: USSN 60/648,957, filed January 31, 2005; USSN 11/344,975, filed January 31, 2006, and PCT application no. US2006/003384, filed January 31, 2006.

The collection of polymorphic sites and the genes in which they occur have a variety of uses. The genes and encoded proteins can be used to identify compounds that modulate the expression or activity of encoded proteins. Such compounds are useful for treatment, prophylaxis, diagnosis or prognosis of AD-related diseases. The collection of genes is also useful for generating transgenic animal models of AD-related disease. These models are useful for screening drugs. The polymorphic sites are also useful in profiling individuals for susceptibility to disease, age-of-onset of disease, response to therapies, or amenability to treatment.

I AD Alzheimer's disease (AD) is a progressive degenerative disease. AD mainly occurs late in life. It is estimated that 2-3 percent of the population over 65 and around 10% of the population over 80 suffer from some form of AD. Roughly half of all AD patients have a positive family history.

Images of brains of patients with AD show significant loss of cells and volume in the regions of the brain devoted to memory and higher mental functioning. Moreover, biopsies of AD patients typically reveal twisted nerve cell fibers, known as neurofibrillary tangles and a sticky protein called beta amyloid.

Neurofibrillary tangles are the damaged remains of microtubules, which allow the flow of nutrients through the neurons (nerve cells). A key component in these tangled fibers is an abnormal form of the tau protein, which in its healthy version helps in the assembly and stabilization of the microtubule structure. The defective tau protein appears to block the actions of the normal version.

Beta amyloid (also called Aβ) is the second significant finding in AD biopsies. This insoluble protein accumulates and forms sticky patches called neuritic plaques, which are found surrounded by the debris of dying nerve cells in the brains of Alzheimer's victims. There are various forms of AD. Early onset AD is a rare form of AD in which people are diagnosed with the disease before age 65. Less than 10% of all AD patients have this type. Early-onset Alzheimer's is strongly hereditary. The hereditary form of early onset AD is also known as familial Alzheimer's disease or FAD. Three genes that have been implicated in early-onset Alzheimer's encode proteins called presenilin 1, presenilin 2, and amyloid precursor protein (APP). The forms of these genes that lead to Alzheimer's are deterministic; virtually everyone who has these forms develops the disease. In other words, the penetrance of certain mutations in these genes is 100%. Each child of a parent who has an Alzheimer-related mutation in one of these genes has a 50 percent chance of inheriting the form that causes Alzheimer's disease. Because of trisomy at chromosome 21, which encodes amyloid precursor protein, people with Down syndrome are particularly at risk of developing a form of early onset AD. Adults with Down syndrome are often in their mid- to late 40s or early 50s when symptoms first appear.

Late-onset AD is the most common form of AD. It usually appears after a person reaches the age of 65. Late-onset AD strikes almost half of all people over the age of 85 and may or may not be hereditary. Late-onset AD is also called sporadic AD. Late-onset Alzheimer's, has a subtler and less clearly understood inheritance pattern. The cholesterol- processing protein called apolipoprotein E (APOE) is a susceptibility gene that occurs in three different alleles: APOE-e4, APOE-e3, and APOE-e2. ApoE-e3 is the most common form and APOE-e2 is the least common. People with one copy of APOE-e4 have an increased chance of developing Alzheimer's, and people with two copies are at even higher risk. However, not everyone with two copies develops Alzheimer's, and many people with the disease have no APOE-e4 at all. In other words, the APOE-e4 allele shows incomplete penetrance. Several other genes identified in the present application also influence the likelihood of developing late-onset Alzheimer's disease.

All embodiments of the invention can be practiced on genes other than APOEl or genomic regions in linkage disequilibrium therewith. However, some embodiments of the invention employ APOEl , other genomic regions in linkage disequilibrium therewith or variant sites in APOEl, particularly in combination with other genes of the invention, as described in more detail below.

II AD Nucleic Acids The invention provides a collection of about 7300 variant sites having forms associated with susceptibility or resistance to AD-related disease, e g , late-onset Alzheimer's disease, or the likelihood of developing an AD-related disease at an early age These variant sites, all of which are SNPs, are provided in Tables A, B, C, D, and/or E (in the files on the CD-R submitted herewith and incorporated herein by reference for all purposes) and all had p values for association of <0 05 The variant sites most associated with AD-related disease occur in the following genes APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4 The p values for the associations of the SNPs in or proximal to these genes was less than 10-7, which indicates a high likelihood that the associations are significant even correcting for all statistical tests performed The other SNPs listed in Tables A, B, C, D, and/or E were identified as nominally significant m their association with AD-related disease Of these genes, APOEl has previously been associated with late-onset Alzheimer's disease Some of the other genes PVRL2, TOMM40, APOCl, APOC4, APOC2, CLPTMl, and BCAM are found on the same chromosome as APOEl Other genes occur at chromosomal locations not previously known to be associated with Alzheimer's disease

PVRL2 or pohovirus receptor-related 2 is also located on chromosome 19, at 19ql3 2-ql3 4, and this gene encodes an adhesion molecule widely expressed in cell lines of different lineages, including hematopoietic, neuronal, endothelial and epithelial cells In 1995, it was identified and reported in humans Later it was identified as a transmembrane molecule related to CDl 55 and named Pohovirus Receptor Related2 (PRR2) It belongs to a family of lmmunoglobulin-hke molecules that includes four members (CDl 11 , CDl 12, PRR3 and CD155) sharing an ectodomam made of three Ig domains, of V, C, C types PVRL2 encodes 2 different transmembrane isoforms sharing identical ectodomams but different transmembrane and cytoplasmic regions The two corresponding transcripts of 4 4 and 3 0 kb were detectable in several tissues PVRL2 is expressed in the myelo-monocytic and megakaryocyte hematopoietic lineages and the function m hematopoiesis is currently unknown PVRL2 is an intercellular homophilic adhesion molecule also known as nectin 2 Homophilic adhesion correlates with the tyrosine phosphorylation of the long isoform PVRL2 localizes specifically at adherens junctions via its cytoplasmic interaction with the scaffold F-actin binding protein afadin This interaction is mediated by a sequence located to the C terminal ends of both isoforms of PVRL2 (A/Ex YV) and the PDZ domain of afadin This sequence is also found in PVRL2 and PRR3 and represents a specific consensus m the family, which is also named nectin family Disruption of the murine PVRL2gene leads to infertility of male mice with morphologically aberrant spermatozoa. PVRL2 mediates entry of some alpha-herpesvirus mutants (also named HveB) via its V domain. PVRL2 is involved in cell to cell spreading of the virus.

TOMM40, located also on chromosome 19, at 19ql3, is a gene thought to be a translocase of outer mitochondrial membrane 40 and used in the import of protein precursors into the mitochondria. Suzuki, H., J Biol Chem. 2000 Dec l;275(48):37930-6. TOMM40 gene products have been found to be expressed in lymph and pancreas cells.

APOE is an important apoprotein of the chylomicron and binds to a specific receptor on liver cells and peripheral cells. APOE is essential for the normal catabolism of triglyceride-rich lipoprotein constituents. The apoE gene is mapped at 19q 13.2 in a cluster with apoCl and apoC2. Defects in APOE result in familial dysbetalipoproteinemia, or type III hyperlipoproteinemia (HLP III), in which increased plasma cholesterol and triglycerides are the consequence of impaired clearance of chylomicron and very low density lipoprotein remnants. APOCl or apolipoprotein Cl, located at chromosome 19 at 19ql3.2, is a protein encoded by a member of the apolipoprotein Cl family. This protein is expressed primarily in the liver, and it is activated when monocytes differentiate into macrophages. A pseudogene of the apocl gene is located 4 kb downstream in the same orientation, on the same chromosome. This gene is mapped to chromosome 19, where it resides within a apolipoprotein gene cluster. Alternatively spliced transcript variants have been found for this gene, but the biological validity of some variants has not been determined.

APOC4 is encoded by apolipoprotein (apo)C4 gene located at 19ql3.2, and which is a member of the apolipoprotein gene family. It is expressed in the liver and has a predicted protein structure characteristic of the other genes in this family. Apo C4 is a 3.3-kb gene consisting of 3 exons and 2 introns; it is located 0.5 kb 5' to the apoC2 gene.

APOC2 is encoded by the apoc2 gene located at 19ql3.2, and is secreted in plasma where it is a component of very low density lipoprotein. This protein activates the enzyme lipoprotein lipase, which hydrolyzes triglycerides and thus provides free fatty acids for cells. Mutations in this gene cause hyperlipoproteinemia type IB, characterized by hypertriglyceridemia, xanthomas, and increased risk of pancreatitis and early atherosclerosis.

CLPTMl is encoded by the CLPTMl gene located at 19ql3.2-ql3.3, and was identified as a novel gene, "cleft lip- and palate-associated transmembrane protein- 1." Assembled cDNA sequences and comparison with genomic sequences predicted a gene with 13 exons encoding a putative protein with 7 transmembrane domains highly conserved between human and C. elegans.

BCAM (also known as "LU") is encoded by the BCAM gene located at 19ql3.2, and is also known as "basal cell adhesion molecule" and "Lutheran blood group glycoprotein." BCAM is a member of the immunoglobulin superfamily and a receptor for the extracellular matrix protein, laminin. The protein contains five, N-terminus, extracellular immunoglobulin domains, a single transmembrane domain, and a short, C-terminal cytoplasmic tail. This protein may play a role in epithelial cell cancer and in vaso-occlusion of red blood cells in sickle cell disease. Two transcript variants encoding different isoforms have been found for this gene. LOC728050 is encoded at 10q23.1, and is a hypothetical protein of unknown function. LOC729099 is encoded at 5pl5.33 and is a hypothetical protein of unknown function.

NEUROG3 (neurogenin 3) is encoded by the NEUROG3 gene located at 10q21.3, and belongs to a family of basic helix-loop-helix transcription factors involved in the determination of neural precursor cells in the neuroectoderm.

Hypothetical protein LOC219738 is encoded by the C10ORF35 open reading frame located at 10q21.3. Its function is unknown, but evidence suggests protein-binding activity and localization to membranes.

The ND3 gene, which is found in the mitochondrial genome, encodes NADH dehydrogenase subunit 3 (ND3, MT-ND3, MTND3), one of seven mitochondrial DNA (mtDNA) encoded subunits (MTNDl, MTND2, MTND3, MTND4, MTND4L, MTND5, MTND6) included among the approximately 41 polypeptides of respiratory Complex I. Complex I accepts electron from NADH, transfers them to ubiquinone (Coenzyme QlO) and uses the energy released to pump protons across the mitochondrial inner membrane. MTND3 has been localized to the hydrophobic protein fragment of the Complex. ND3 may be an important factor in Parkinson's disease susceptibility among white individuals and could help explain the role of Complex I in Parkinson disease expression. There is also some evidence to suggest a role in Leigh syndrome and dystonia.

The ND4 gene, which is found in the mitochondrial genome, encodes NADH dehydrogenase subunit 4 (ND4, MT-ND4, MTND4), which (like ND3) is one of seven mitochondrial DNA (mtDNA) encoded subunits of respiratory Complex I. MTND4 is probably a component of the hydrophobic protein fragment. MTND4 mutations have been implicated in Leber hereditary optic neuropathy, Leber optic atrophy and dystonia, Wolfram syndrome, and MELAS syndrome. Table A in the file named "filtered_all_pO5_latest_v2.txt" on the CD-R incorporated herein by reference for all purposes contains SNPs identified as associated with AD-related disease in a two-step association study consisting of a first pooled stage followed by an individual genotyping stage to validate the SNPs identified in the pooled stage. Approximately 250,000 "tag" SNPs were analyzed by pooled genotyping in a first sample set, consisting of the "original" case and control samples. Approximately 20,000 SNPs were identified as "possibly associated" with the phenotype and were reanalyzed by individual genotyping in the original case and control samples, as well as additional "replication" case and control samples. As such, the "original" case and control samples are defined as those used hi the both the pooled and individual genotyping stages of the study and are sometimes referred to as "pg" samples; and the "replication" case and control samples are defined as those used only in the individual genotyping stage, and are sometimes referred to as "nonpooled genotyping" or "npg" samples.

Table A, column 1, entitled "REFSNPJD," provides the rsID assigned to the SNP position from National Center for Biotechnology Information (NCBI; ncbi.nlm.nih.gov), if available.

Table A, column 2, entitled "SNP_ID," is an internal Perlegen number that identifies a single variant position.

Table A, column 3, entitled "ACCESSION_NUM," provides the accession number from NCBI Build 36.2 of the contig to which the SNP aligns, if available.

Table A, column 4, entitled "POSITION," provides the nucleotide position in the NCBI Build 36.2 contig of the "N" position in the assayed sequence provided in Table C, if available.

Table A, column 5, entitled "DELTA_P," identifies the difference in allele frequency of Allele 1 expressed as Allele 1 frequency in cases minus Allele 1 frequency in controls. A positive value in this column means that Allele 2 is found in excess in controls (i.e., is protective for AD-related disease, e.g. Alzheimer's disease) and, therefore, that an individual carrying Allele 1 is at a higher risk of AD-related disease. A negative value in this column indicates that Allele 1 is found in excess in controls, suggesting that an individual carrying Allele 2 is at a higher risk of AD-related disease. These analyses are described more fully in Example 18, herein.

Table A, column 6, entitled "IG_GLM_P_VALUE," identifies the p-value of a chi- square statistic, representing the significance of the difference in a logistic regression comparing a model with genotype terms to one without. This statistic is computed over the original case and control samples

Table A, column 7, entitled "IG GLM P FDR," identifies the false discovery rate for the test on original case and control samples

Table A, column 8, entitled "REP_GLM_P_VALUE," identifies p-value of a chi- square statistic, representing the significance of the difference in a logistic regression comparing a model with genotype terms to one without This statistic is computed over the replication case and control samples

Table A, column 9, entitled "REP_GLM_P_FDR," identifies the false discovery rate for the test on replication case and control samples Table A, column 10, entitled "ALL_GLM_P_VALUE," identifies p-value of a chi- square statistic, representing the significance of the difference in a logistic regression comparing a model with genotype terms to one without This statistic is computed over all case and control samples

Table A, column 11, entitled "ALL_GLM_P_FDR," identifies the false discovery rate for the test on all case and control samples

Table A, column 12, entitled "NEARB Y GENES," identifies genes that contain or flank the SNP within 50 kb, and at least one upstream and downstream gene 1 Mb for mapped SNPs A spacer "— " indicates an interval of more than 50 kb, and longer spacers indicate longer intervals Genes that contain the SNP are enclosed in square brackets Table A, column 13, entitled "GENEJN AME" provides the NCBI symbol for a gene withm 10 kb of the SNP position

Table A, column 14, entitled "HITJTYPE," identifies where the SNP lies within or relative to a gene A "HITJTYPE" of "intron" means that the SNP occurs within an intron of a gene A "HITJTYPE" of "exon" means that the SNP occurs within an exon of a gene A "HITJTYPE" of "down" means that the SNP occurs within 10 kb downstream of a gene A "HITJTYPE" of "up" means that the SNP occurs within 10 kb upstream from the start codon Table A, column 15, entitled "SYNONYMOUS," indicates whether the SNP alleles code for different amino acids in an encoded protein "yes" indicates the two alleles encode the same protein sequence, and "no" indicates that the two alleles encode different amino acids at the same position in the resulting protein "outsideCodmgRegion" indicates the SNP occurs outside the coding region of a gene "unknown_poor_alignment_at_snp_pos" indicates the synonymous nature of the SNP alleles cannot be determined due to poor local alignment There is no entry for those SNP positions that do not occur within an exon

Table A, column 16, entitled "REFSEQ_AA," identifies the reference amino acid at the SNP position, if any This can have a value of "X" if the ammo acid is unknown or if there are ambiguity characters in the accessioned mRNA sequence

Table A, column 17, entitled "REF_AA," provides the alternate amino acid residue if SNP Allele 1 causes a change in the amino acid sequence with respect to the reference ammo acid

Table A, column 18, entitled "ALT_AA," provides the alternate ammo acid residue if SNP Allele 2 causes a change in the amino acid sequence with respect to the reference ammo acid

Thus, Table A illustrates variant sites and associated gene regions having variant forms associated with resistance or susceptibility to AD-related disease Such associated gene regions include APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, AP0C4, LOC728050, NEUROG3, C10ORF35, ND3, and ND4 and any fragments or derivatives thereof Optionally, the group can further comprise other genes shown in Table A, e g , BCAM, PSEN2, APP, HDAC4, OLFMl, RAB 12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B. EXOC2, and LOC642335 Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, AP0C4, APP, or PSEN2

Table B in the file named "filtered_40416_SNPs txt" on the CD-R incorporated herein by reference for all purposes contains SNPs identified as associated with AD-related disease in another two-step association study consisting of a first pooled stage followed by an individual genotypmg stage to validate the SNPs identified in the pooled stage

Approximately 1 6 million SNPs were analyzed by pooled genotypmg in a first sample set, consisting of the "original" case and control samples Approximately 40,000 SNPs were identified as "associated" with the phenotype and were reanalyzed by individual genotypmg in the original case and control samples, as well as additional "replication" case and control samples As such, the "original" case and control samples are defined as those used in the both the pooled and individual genotypmg stages of the study and are sometimes referred to as "pg" samples, and the "replication" case and control samples are defined as those used only m the individual genotypmg stage, and are sometimes referred to as "nonpooled genotypmg" or "npg" samples Table B, column 1 , entitled "REFSNP ID," provides the rsID assigned to the SNP position from National Center for Biotechnology Information (NCBI, ncbi nlm mh gov), if available

Table B, column 2, entitled "SNP_ID," is an internal Perlegen number that identifies a single variant position Table B, column 3, entitled "ACCESSIONJNUM," provides the accession number from NCBI Build 36.2 of the contig to which the SNP aligns, if available.

Table B, column 4, entitled "POSITION," provides the nucleotide position in the NCBI Build 36.2 contig of the "N" position in the assayed sequence provided in Table C, if available.

Table B, column 5, entitled "DELTA_P," identifies the difference in allele frequency of Allele 1 expressed as Allele 1 frequency in cases minus Allele 1 frequency in controls. A positive value in this column means that Allele 2 is found in excess in controls (i.e., is protective for AD-related disease, e.g. Alzheimer's disease) and, therefore, that an individual carrying Allele 1 is at a higher risk of AD-related disease. A negative value in this column indicates that Allele 1 is found in excess in controls, suggesting that an individual carrying Allele 2 is at a higher risk of AD-related disease. These analyses are described more fully in Example 20, herein.

Table B, column 6, entitled "PG_P_VALUE," identifies the p-value of a trend score statistic. This statistic is computed over the original case and control samples.

Table B, column 7, entitled "NPG_P_VALUE," identifies the p-value of a trend score statistic. This statistic is computed over the replication case and control samples.

Table B, column 8, entitled "PGNPG P V ALUE," identifies the p-value of a trend score statistic. This statistic is computed over all case and control samples. Table B, column 9, entitled "NEARB Y_GENES," identifies genes that contain or flank the SNP within 50 kb, and at least one upstream and downstream gene 1 Mb for mapped SNPs. A spacer "—" indicates an interval of more than 50 kb, and longer spacers indicate longer intervals. Genes that contain the SNP are enclosed in square brackets.

Table B, column 10, entitled "GENE_NAME" provides the NCBI symbol for a gene within 10 kb of the SNP position.

Table B, column 11 , entitled "HITJTYPE," identifies where the SNP lies within or relative to a gene. A "HIT-TYPE" of "intron" means that the SNP occurs within an intron of a gene. A "HITJTYPE" of "exon" means that the SNP occurs within an exon of a gene. A "HITJTYPE" of "down" means that the SNP occurs within 10 kb downstream of a gene. A "HITJTYPE" of "up" means that the SNP occurs within 10 kb upstream from the start codon.

Table B, column 12, entitled "SYNONYMOUS," indicates whether the SNP alleles code for different amino acids in an encoded protein, "yes" indicates the two alleles encode the same protein sequence, and "no" indicates that the two alleles encode different amino acids at the same position in the resulting protein. "outsideCodingRegion" indicates the SNP occurs outside the coding region of a gene. "unknown_poor_alignment_at_snp_pos" indicates the synonymous nature of the SNP alleles cannot be determined due to poor local alignment. There is no entry for those SNP positions that do not occur within an exon.

Table B, column 13, entitled "REFSEQ AA," identifies the reference amino acid at the SNP position, if any. This can have a value of "X" if the amino acid is unknown or if there are ambiguity characters in the accessioned mRNA sequence.

Table B, column 14, entitled "REF_AA," provides the alternate amino acid residue if SNP Allele 1 causes a change in the amino acid sequence with respect to the reference amino acid. Table B, column 15, entitled "ALT_AA," provides the alternate amino acid residue if SNP Allele 2 causes a change in the amino acid sequence with respect to the reference amino acid.

Thus, Table B illustrates variant sites and associated gene regions having variant forms associated with resistance or susceptibility to AD-related disease. Such associated gene regions include: APOE, APOCl, PVRL2, BCAM, TOMM40, CLPTMl, APOC2, APOC4, and LOC729099, and any fragments or derivatives thereof. Optionally, the group can further comprise other genes shown in Table B, e.g., HDAC4, LOC389300, ZNF366, LOC644154, and LOC645932. Optionally, the group can exclude APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, APP, or PSEN2. Table C in the file named "EvsL_Logistic_l-5-07.txt" on the CD-R incorporated herein by reference for all purposes contains information pertaining to the use of logistic regression models for analysis of association study data to find SNPs associated with age of onset of Alzheimer's disease. All samples were Caucasian Alzheimer's patients. After samples with very early ages of onset (<50) were removed, samples in the lower and upper 25% of the age of onset distribution were used as cases and controls, respectively. Pooled genotyping was carried out before SNPs were selected for individual genotyping. Further description of this analysis is provided in Example 23.

Table C, column 1, entitled "REFSNPJD," provides the rsID assigned to the SNP position from National Center for Biotechnology Information (NCBI; ncbi.nlm.nih.gov), if available.

Table C, column 2, entitled "SNP_ID," is an internal Perlegen number that identifies a single variant position.

Table C, column 3, entitled "ACCESSION_NUM," provides the accession number from NCBI Build 36.2 of the contig to which the SNP aligns, if available. Table C, column 4, entitled "POSITION," provides the nucleotide position in the NCBI Build 36.2 contig of the "N" position in the assayed sequence provided in Table C, if available.

Table C, column 5, entitled "PVAL," identifies the p-value of a chi-square statistic obtained for each SNP by testing the following two nested logistic regression models: early_onset ~ PCl , and early onset ~ PCl + genotype, where genotype is coded as 0, 1, or 2.

Table C, column 6, entitled "EFFECT," identifies the coefficient of the genotype term in the model. A positive value in this column means that Allele 1 is found at a higher frequency in earlier age-of-onset cases (i.e., Allele 1 is associated with early onset of the disease). A negative value in this column indicates that Allele 2 is found at a higher frequency in earlier age-of-onset cases (i.e., Allele 2 is associated with early onset of the disease). These analyses are described more fully in Example 23, herein. Table C, column 7, entitled "NEARB Y_GENES," identifies genes that contain or flank the SNP within 50 kb, and at least one upstream and downstream gene 1 Mb for mapped SNPs. A spacer "— " indicates an interval of more than 50 kb, and longer spacers indicate longer intervals. Genes that contain the SNP are enclosed in square brackets.

Table C, column 8, entitled "GENE-NAME" provides the NCBI symbol for a gene within 10 kb of the SNP position.

Table C, column 9, entitled "HIT_TYPE," identifies where the SNP lies within or relative to a gene. A "HIT TYPE" of "intron" means that the SNP occurrs within an intron of a gene. A "HIT_TYPE" of "exon" means that the SNP occurs within an exon of a gene. A "HIT_TYPE" of "down" means that the SNP occurs within 10 kb downstream of a gene. A "HITJTYPE" of "up" means that the SNP occurs within 10 kb upstream from the start codon. Table C, column 10, entitled "SYNONYMOUS," indicates whether the SNP alleles code for different amino acids in an encoded protein, "yes" indicates the two alleles encode the same protein sequence, and "no" indicates that the two alleles encode different amino acids at the same position in the resulting protein. "outsideCodingRegion" indicates the SNP occurs outside the coding region of a gene. "unknown_poor_alignment_at_snp _pos" indicates the synonymous nature of the SNP alleles cannot be determined due to poor local alignment. There is no entry for those SNP positions that do not occur within an exon.

Table C, column 11, entitled "REFSEQ-AA," identifies the reference amino acid at the SNP position, if any. This can have a value of "X" if the amino acid is unknown or if there are ambiguity characters in the accessioned mRNA sequence.

Table C, column 12, entitled "REF_AA," provides the alternate amino acid residue if SNP Allele 1 causes a change in the amino acid sequence with respect to the reference amino acid. Table C, column 13, entitled "ALT_AA," provides the alternate amino acid residue if SNP Allele 2 causes a change in the amino acid sequence with respect to the reference amino acid.

Table D in the file named "AOO_lm_l-5-07.txt" on the CD-R incorporated herein by reference for all purposes contains information pertaining to the use of linear regression models for analysis of association study data to find SNPs associated with age of onset of Alzheimer's disease. All samples were Caucasian Alzheimer's patients. After samples with very early ages of onset (<50) were removed, samples in the lower and upper 25% of the age of onset distribution were used as cases and controls, respectively. Pooled genotyping was carried out before SNPs were selected for individual genotyping. Further description of this analysis is provided in Example 23.

Table D, column 1, entitled "REFSNPJD," provides the rsID assigned to the SNP position from National Center for Biotechnology Information (NCBI; ncbi.nlm.nih.gov), if available.

Table D, column 2, entitled "SNP_ID," is an internal Perlegen number that identifies a single variant position.

Table D, column 3, entitled "ACCES SI0N_NUM," provides the accession number from NCBI Build 36.2 of the contig to which the SNP aligns, if available.

Table D, column 4, entitled "POSITION," provides the nucleotide position in the NCBI Build 36.2 contig of the "N" position in the assayed sequence provided in Table E, if available.

Table D, column 5, entitled "PVAL," identifies the p-value of an F statistic obtained for each SNP by testing the following two nested linear regression models: age_of_onset ~ PCl, and age_of_onset ~ PCl + genotype, where genotype is coded as 0, 1, or 2; and age is coded as 0 for late onset and 1 for early onset.

Table D, column 6, entitled "EFFECT," identifies the coefficient of the genotype term in the model. A positive value in this column means that Allele 2 is found at a higher frequency in earlier age-of-onset cases (i.e., Allele 2 is associated with early onset of the disease) A negative value in this column indicates that Allele 1 is found at a higher frequency m earlier age-of-onset cases (l e , Allele 1 is associated with early onset of the disease) These analyses are described more fully in Example 23, herein

Table D, column 7, entitled "NEARBY_GENES," identifies genes that contain or flank the SNP within 50 kb, and at least one upstream and downstream gene 1 Mb for mapped SNPs A spacer "— " indicates an interval of more than 50 kb, and longer spacers indicate longer intervals Genes that contain the SNP are enclosed in square brackets

Table D, column 8, entitled "GENE NAME" provides the NCBI symbol for a gene withm 10 kb of the SNP position Table D, column 9, entitled "HIT TYPE," identifies where the SNP lies within or relative to a gene A "HITJTYPE" of "intron" means that the SNP occurs within an nitron of a gene A "HIT_T YPE" of "exon" means that the SNP occurs within an exon of a gene A "HITJTYPE" of "down" means that the SNP occurs withm 10 kb downstream of a gene A "HITJTYPE" of "up" means that the SNP occurs within 10 kb upstream from the start codon Table D, column 10, entitled "SYNONYMOUS," indicates whether the SNP alleles code for different ammo acids in an encoded protein "yes" indicates the two alleles encode the same protein sequence, and "no" indicates that the two alleles encode different amino acids at the same position m the resulting protein "outsideCodingRegion" indicates the SNP occurs outside the coding region of a gene "unknown_poor_alignment_at_snp_pos" indicates the synonymous nature of the SNP alleles cannot be determined due to poor local alignment There is no entry for those SNP positions that do not occur withm an exon

Table D, column 11, entitled "REFSEQ AA," identifies the reference amino acid at the SNP position, if any This can have a value of "X" if the ammo acid is unknown or if there are ambiguity characters m the accessioned mRNA sequence Table D, column 12, entitled "REF_AA," provides the alternate ammo acid residue if SNP Allele 1 causes a change in the amino acid sequence with respect to the reference amino acid

Table D, column 13, entitled "ALT_AA," provides the alternate ammo acid residue if SNP Allele 2 causes a change m the amino acid sequence with respect to the reference amino acid

Table E in the file named "SNP information txt" on the CD-R incorporated herein by reference for all purposes contains additional information pertaining to the variant sites shown in Tables A, B, C, and D having variant forms associated with resistance or susceptibility to AD-related disease Table E, column 1, entitled "SNP_ID", provides an internal Perlegen number that identifies a single variant position. This same numbering is used in Tables A and B.

Table E, column 2, entitled "dbSNP_rsID" provides the rsID assigned to the SNP position from NCBI, if available. This same numbering is used in the columns entitled "REFSNPJD" in Tables A and B.

Table E, column 3, entitled "dbSNP_ssID" provides the ssID assigned to the SNP that Perlegen submitted to dbSNP, if available. If a SNP has an rsID but not an ssID, this means that Perlegen has not submitted this SNP to dbSNP, but an existing SNP in dbSNP maps (in the Perlegen alignment process) to the same location as the Perlegen SNP. Table E, column 4, entitled "Allele 1" provides the nucleotide code for Perlegen' s reference alleles and Table E, column 5, entitled "Allele 2" provides the nucleotide code for Perlegen' s alternate alleles. The designation of reference and alternate alleles is arbitrary as far as resistance or susceptibility to AD is concerned. However, such can be determined using the information provided in Tables A, B, D, or E, as described herein. Table E, column 6, entitled "Chromosome" provides the chromosome number of the

NCBI Build 36.2 contig on which the best alignment was found. X symbolizes the X chromosome. Y symbolizes the Y chromosome. U symbolizes sequences not assigned to any chromosome on Build 36.2. This field may be null if a SNP could not be placed on any contig. Table E, column 7, entitled "sex-linked" provides information on sex-linkage of the

SNP. "A" represents an autosomal SNP; "P" represents a pseudoautosomal SNP (e.g., on X or Y chromosomes in the pseudoautosomal region); "S" represents a sex-linked SNP (either on the X or Y chromosome, but not in the pseudoautosomal region); and U represents an unassigned (or unknown pseudoautosomal status for X and Y). Table E, column 8, entitled "Accession ID" represents the accession number from

NCBI Build 36.2 of the contig to which the SNP aligns. This may be null.

Table E, column 9, entitled "Contig Position" represents the nucleotide position in NCBI Build 36.2 contig of the "N" position in the assayed sequence in column 11. For SNPs this is always a single position, but in the case of a deletion-insertion polymorphism (DIP), the mapping may be a range. This may be null.

Table E, column 10, entitled "Strand" is a + or -, based on the strand for Allele 1 on NCBI Build 36.2. This may be null.

Table E, column 11 , entitled "Assayed sequence" is the 29-mer (SNPs) or 30-mer (DIPs) that was used to assay the SNP on a microarray, with an ambiguity character "N" representing the SNP or DIP at the middle base.

Additional variants (and their associated gene regions) that can be used to diagnose, treat, or prevent AD-related disease include, but are not limited to, those in linkage disequilibrium with those identified in Tables A, B, C, D, and/or E, e.g., in haplotype blocks with the variants identified in Tables A, B, C, D, and/or E. Such variants can be identified according to, e.g., U.S. Ser. No. 10/106,097, entitled "Methods For Genomic Analysis", filed March 26, 2002; U.S. Ser. No. 10/284,444, filed October 31, 2002, entitled "Human Genomic Polymorphisms"; and Patil, N. et al, "Blocks of Limited Haplotype Diversity Revealed by High-Resolution Scanning of Human Chromosome 21" Science 294, 1719-1723 (2001). A variant in linkage disequilibrium with a variant of Tables A, B, C, D, and/or E that is associated with AD-related disease is also associated with AD-related disease. For example, a variant in a haplotype block with a variant of Tables A, B, C, D, and/or E that is associated with AD-related disease is also associated with AD-related disease. More specifically, an allele of a variant in a haplotype pattern with an allele of a variant of Tables A, B, C, D, and/or E that is associated with resistance to AD-related disease is also associated with resistance to AD-related disease. Similarly, an allele of a variant in a haplotype pattern with an allele of a variant associated with susceptibility to AD-related disease identified in Tables A, B, C, D, and/or E is also associated with susceptibility to AD-related disease.

The genes showing the strongest associations with resistance or susceptibility to Alzheimer's disease, or the likelihood of developing an AD-related disease at an early age, are shown in the following table:

The preferred polymorphic sites in or around these genes and other loci associated with resistance/susceptibility or age-of-onset of AD-related disease are shown, e.g., in Tables A, B, C, and D. The polymorphisms, alleles and associated genomic regions identified herein can be used to identify, isolate and amplify nucleic acids associated with resistance or susceptibility (including early age-of-onset) to AD and AD-related disease. Such nucleic acids can be used for prognostics, diagnostics, theranostics, prevention, treatment and further study of AD and AD-related disease. For example, in one embodiment, an AD nucleic acid is a nucleotide sequence from the human genome that comprises a nucleic acid in a position identified in Tables A, B, C, D, and/or E, or a nucleic acid in linkage disequilibrium with a nucleic acid position identified in Tables A, B, C, D, and/or E, or a nucleic acid in haplotype block or pattern with a nucleic acid position identified in Tables A, B, C, D, and/or E. Such AD nucleic acids can include coding sequence and/or non-coding sequences. They can comprise, consist essentially of, or consist of one or more exons or introns encompassing such nucleic acid positions. They can be of variable length. In some embodiments such nucleic acids can be less than 500,000, 100,000, 50,000, 10,000, 5,000, 1,000, 500, 100, 10 or 5 nucleotides in length. In some embodiments such nucleic acids can be greater than 5, 10, 50, 100, 300, 600, 900, 1,000, 3,000, 6,000, 9,000, 10,000, 30,000, 60,000, 90,000, 100,000, 300,000, 600,000, or 900,000 nucleotides in length.

In one embodiment, an AD nucleic acid is one that can specifically hybridizes to an associated genomic region encompassing a nucleic acid position identified in Tables A, B, C, D, and/or E, or an associated genomic region comprising a nucleic acid in linkage disequilibrium with a position identified in Tables A, B, C, D, and/or E, or an associated genomic region comprising a nucleic acid in a haplotype block or pattern with a position identified in Tables A, B, C, D, and/or E. In one embodiment, nucleic acids disclosed herein that can specifically hybridize to a genomic region associated with an AD-related disease, are identified in Table E. Due to the duplex nature of DNA, sequences complementary to those provided in Table E can also specifically hybridize to a genomic region associated with PD- related disease and are contemplated to be part of the instant invention. Thus, nucleic acids provided herein or complementary sequences thereto can, in some embodiments, specifically hybridize to a genomic sequence having one or more polymorphisms identified in Tables A, B, C, D, and/or E and/or other polymorphisms in linkage disequilibrium with (e g , m the same haplotype blocks as) the polymorphisms in Tables A, B, C, D, and/or E Methods for identifying polymorphisms in a haplotype block and in haplotype patterns within a haplotype block are provided in U S Ser No 10/106,097 entitled "Methods For Genomic Analysis," filed March 26, 2002, U S Ser No 10/284,444, filed October 31, 2002, entitled "Human Genomic Polymorphisms", in US 20040023275, in U S patent application no 10/367,558, filed February 14, 2003, entitled "Identifying SNP Patterns" (all of which are assigned to the same assignee as the present application), and in Patil, et al (2001) "Blocks of Limited Haplotype Diversity Revealed by High-Resolution Scanning of Human Chromosome 21", Science 294 1719-1723

In some embodiments, the nucleic acids herein are associated with AD or age-of- onset thereof Nucleic acids associated with resistance to AD comprise at least one allele associated with resistance to AD or an allele in linkage disequilibrium with (e g , in a haplotype pattern with) an allele associated with resistance to AD Nucleic acids associated with susceptibility to AD comprise at least one allele associated with susceptibility to AD or an allele in linkage disequilibrium with (e g , in a haplotype pattern with) an allele associated with susceptibility to AD In some embodiments, a nucleic acid associated with resistance to AD is one that is expressed differently in individuals having a phenotype of resistance to AD as compared to individuals having who do not have a phenotype of resistance to AD, or a nucleic acid having one or more alleles associated with resistance to AD For example, a nucleic acid associated with resistance to AD is one that can specifically hybridize to a genomic region having one or more alleles of polymorphisms identified in Tables A, B, C, D and/or E as being associated with resistance to AD, or one or more alleles in linkage disequilibrium therewith (e g , in a haplotype pattern therewith) In other embodiments, a nucleic acid associated with susceptibility to AD is one that is expressed differently in individuals having a phenotype of susceptibility to AD as compared to individuals having who do not have a phenotype of susceptibility to AD, or a nucleic acid having one or more alleles associated with susceptibility to AD For example, a nucleic acid associated with susceptibility to AD is one that can specifically hybridize to a genomic region having one or more alleles of polymorphisms identified in Tables A, B, C, D and/or E as being associated with susceptibility to AD, or one or more alleles m linkage disequilibrium therewith (e g , in a haplotype pattern therewith) In certain embodiments, a nucleic acid associated with susceptibility to AD is associated with an earlier age-of-onset of AD, and a nucleic acid associated with resistance to AD is associated with a later age-of-onset of AD.

In certain embodiments, a set of nucleic acids is provided that can specifically hybridize to at least 2 polymorphisms, preferably at least 3 polymorphisms, at least 4 polymorphisms, at least 5 polymorphisms, at least 6 polymorphisms, at least 7 polymorphisms, at least 8 polymorphisms, or at least 9 polymorphisms associated with AD- related disease such as those identified in Tables A, B, C, D and/or E, and/or polymorphisms in linkage disequilibrium therewith (e.g., in haplotype blocks therewith), or complementary sequences thereto. In other embodiments, a set of nucleic acids is provided that can specifically hybridize to at least 2 alleles, preferably at least 3 alleles, at least 4 alleles, at least 5 alleles, at least 6 alleles, at least 7 alleles, at least 8 alleles, or at least 9 alleles associated with resistance to AD-related disease, and/or alleles in linkage disequilibrium therewith (e.g., in haplotype patterns therewith), or complementary sequences thereto. Similarly, a set of nucleic acids may be provided that can specifically hybridize to at least 2 alleles, preferably at least 3 alleles, at least 4 alleles, at least 5 alleles, at least 6 alleles, at least 7 alleles, at least 8 alleles, or at least 9 alleles associated with susceptibility to AD-related disease, and/or alleles in linkage disequilibrium therewith (e.g., in haplotype patterns therewith), or complementary sequences thereto.

A nucleic acid can be single-stranded or double-stranded. It can also comprise coding (e.g., exon) or non-coding sequence (e.g., introns, 3' or 5' untranslated regions, and regulatory regions) or a combination of coding and non-coding nucleic acids. In a preferred embodiment, a coding AD nucleic acid is one that can specifically hybridize to at least a portion of the coding region of an associated gene, or to one or more exons of an associated gene, or to one or more open reading frames of an associated gene.

A nucleic acid provided herein can be fused to at least one other nucleic acid (e.g., a tag sequence or reporter gene) to create a construct for producing a specific protein product, such as a fusion protein. A tag sequence encodes a polypeptide that can assist in isolation or purification of the protein product (e.g., glutathione S transferase (GST) fusion protein or a hemagglutinin A (HA) polypeptide). A reporter gene also encodes an easily assayed protein and is often used to replace other coding regions whose protein products are difficult to assay. A fusion protein is formed by the expression of a hybrid nucleic acid made by combining two nucleic acid sequences.

Conditions for nucleic acid hybridization vary depending on the buffers used, length of nucleic acids, ionic strength, temperature, etc. The term "stringent conditions" for hybridization refers to the incubation and wash conditions (e.g., conditions of temperature and buffer concentration) that permit hybridization of a first nucleic acid to a second nucleic acid. The first nucleic acid may be perfectly (e.g. 100%) complementary to the second or may share some degree of complementarity, which is less than perfect (e.g., more than 70%, 75%, 85%, or 95%). For example, certain high stringency conditions can be used which distinguish perfectly complementary nucleic acids from those less complementary, even those having only a single base mismatch. High stringency, moderate stringency and low stringency conditions for nucleic acid hybridization are known in the art. Ausubel, F.M. et al., "Current Protocols in Molecular Biology" (John Wiley & Sons 1998), pages 2.10.1- 2.10.16; 6.3.1-6.3.6. The exact conditions which determine the stringency of hybridization depend not only on ionic strength (e.g., 0.2XSSC, 0. IXSSC), temperature (e.g., room temperature, 42°C, 68°C) and the concentration of destabilizing agents such as formamide or denaturing agents such as SDS, but also on factors such as the length of the nucleic acid sequence, base composition, percent mismatch between hybridizing sequences and the frequency of occurrence of subsets of that sequence within other non-identical sequences. Thus, equivalent conditions can be determined by varying one or more of these parameters while maintaining a similar degree of identity or similarity between the two nucleic acid molecules. Typically, conditions are used such that sequences at least about 60%, at least about 70%, at least about 80%, at least about 90% or at least about 95% or more identical to each other remain hybridized to one another. By varying hybridization conditions from a level of stringency at which no hybridization occurs to a level at which hybridization is observed, conditions which will allow a given sequence to hybridize (e.g., selectively) with the most similar sequences in the sample can be determined. Exemplary conditions are described in Krause, et al., Methods in Enzymology, (1991) 200:546-556 and in Ausubel, et al., "Current Protocols in Molecular Biology", (John Wiley & Sons 1998), which describes the determination of washing conditions for moderate or low stringency conditions. Washing is the step in which conditions are usually set so as to determine a minimum level of complementarity of the hybrids. Generally, starting from the lowest temperature at which only homologous hybridization occurs, each °C by which the final wash temperature is reduced (holding SSC concentration constant) allows an increase by 1% in the maximum extent of mismatches among the sequences that hybridize. Generally, doubling the concentration of SSC results in an increase in TM of ~17°C. Using these guidelines, the washing temperature can be determined empirically for high, moderate or low stringency, depending on the level of mismatch sought. For example, a low stringency wash can comprise washing in a solution containing 0.2XSSC/0.1% SDS for 10 min at room temperature; a moderate stringency wash can comprise washing in a prewarmed solution (42°C) solution containing 0.2XSSC/0.1% SDS for 15 min at 42⁰C; and a high stringency wash can comprise washing in prewarmed (68⁰C) solution containing O.lXSSC/0.1 %SDS for 15 min at 68°C. Furthermore, washes can be performed repeatedly or sequentially to obtain a desired result as known in the art. Equivalent conditions can be determined by varying one or more of the parameters given as an example, as known in the art, while maintaining a similar degree of identity or similarity between the target nucleic acid and the primer or probe used. Additional descriptions of nucleic acid hybridization techniques are provided, e.g., in U.S.S.N. 11/058,432, filed February 14, 2005; U.S.S.N. 11/173,309, filed June 30, 2005; and U.S.S.N. 61/000,752, filed October 26, 2007.

Furthermore, a nucleic acid is preferably isolated. Various nucleic acid isolation techniques are well known in the art, such as those described in Sambrook, et al., Molecular Cloning: A Laboratory Manual (Cold Spring Harbor Laboratory, New York) (1989), and Ausubel, et al., Current Protocols in Molecular Biology (John Wiley and Sons, New York) (1997). For example, an isolated nucleic acid is one that is separated from the nucleic acids that normally flank it or from other biological materials (e.g., other nucleic acids, proteins, lipids, cellular components, etc.) in a sample.

Nucleic acids may also be amplified using polymerase chain reaction (PCR) and other techniques known in the art. See Erlich, H.A., "PCR Technology: Principles and Applications for DNA Amplification" (ed. Freeman Press, NY, NY, 1992); Innis M.A., et al., "PCR Protocols: A Guide to Methods and Applications" (Eds. Academic Press, San Diego, CA, 1990). In addition to PCR, other suitable isolation and amplification methods include, for example, the ligase chain reaction (LCR) (see Wu and Wallace, Genomics, 4:560 (1989), Landegren et al., Science, 241 :1077 (1988), transcription amplification (Kwoh et al., Proc. Natl. Acad. Sci. USA, 86 : 1173 (1989)), self-sustained sequence replication (Guatelli et al., Proc. Natl. Acad. Sci. USA, 87:1874 (1990)) and nucleic acid based sequence amplification (NASBA). The latter two amplification methods involve isothermal reactions based on isothermal transcription that produces both single-stranded RNA (ssRNA) and double- stranded DNA (dsDNA) as the amplified products in a ratio of approximately 30-100 fold more ssRNA than dsDNA. Certain methods for primer selection and amplification are detailed in, e.g., U.S. Patent No. 6,898,531; U.S. Patent No. 6,740,510; U.S.S.N. 10/236,480, filed September 5, 2002; and U.S.S.N. 10/341,832, filed January 14, 2003. Amplification methods may result in a subset of sequences being selected from a complex sample, such as those described in U.S.S.N. 11/058,432, filed February 14, 2005; and U.S.S.N. 61/000,752, filed October 26, 2007, both of which are entitled "Selection Probe Amplification".

Further, homologues of the AD nucleic acids presented herein may be present in other species, and may be identified and readily isolated without undue experimentation by molecular biological techniques well known in the art using the polymorphisms, alleles and associated genomic regions identified herein. Further, there may exist nucleic acids at other locations within the genome that encode proteins that have extensive homology to one or more domains of the AD polypeptides herein. These nucleic acids may be identified via similar techniques.

For example, an AD nucleic acid may be labeled and used to screen a genomic or cDNA library constructed from mKNA obtained from the organism of interest. Hybridization conditions will be of a lower stringency when the cDNA library was derived from an organism different from the type of organism from which the labeled nucleic acid was derived. Such lower stringency conditions vary predictably depending on the specific organisms from which the library and the labeled nucleic acids are derived. For guidance regarding such conditions see, for example, Sambrook et al. (1989) Molecular Cloning, A Laboratory Manual, Cold Spring Harbor Press, N. Y.; and Ausubel et al. (1989) Current Protocols in Molecular Biology, Green Publishing Associates and Wiley Interscience, N. Y. 1. Probes and Primers

The nucleic acids herein can be used as probes and primers in various assays. The terms "probe(s)" and "primer(s)" refer to nucleic acids that hybridize, in whole or in part, in a sequence-specific manner to a complementary strand. Probes and primers include polypeptide nucleic acids, such as those described in Nielsen et al. (1991) Science 254:1497- 1500.

In certain embodiments, the term "primer" refers to a single-stranded nucleic acid that can act as a point of initiation of template-directed DNA synthesis, such as in PCR. PCR reactions can be designed based on the human genome sequence and the associated genomic regions or polymorphisms identified in Tables A, B, C, D, and/or E. For example, where a polymorphism is located in an exon, the exon can be isolated and amplified using primers that are complementary to the nucleotide sequences at both ends of the exon. Similarly, where a polymorphism is located in an intron, the entire intron can be isolated and amplified using primers that are complementary to the nucleotide sequences at both ends of the intron. Short- or long-range PCR primers may be designed to amplify the associated genomic regions or polymorphisms identified in Tables A, B, C, D, and/or E using methods known in the art and further described in U.S. Patent No. 6,898,531; U.S. Patent No. 6,740,510, and U.S.S.N. 10/341,832, filed January 14, 2003, entitled "Apparatus and Methods for Selecting PCR Primer Pairs".

In some embodiments, a probe or primer contains a region of at least about 10 contiguous nucleotides, preferably at least about 15 contiguous nucleotides, more preferably about 20 or about 30 or about 50 contiguous nucleotides, that can specifically hybridize to a complementary nucleic acid sequence. In addition, a probe or primer is preferably about 100 or fewer nucleotides, more preferably between 6 and 50 nucleotides, and more preferably between 12 and 30 nucleotides in length. In certain embodiments, a first portion of a probe or primer is perfectly complementary to a target nucleic acid, and a second portion of the probe or primer is not perfectly complementary to the target nucleic acid. In some aspects, the portion that is not perfectly complementary contains a binding site, e.g., for a polypeptide or another probe or primer.

To isolate, amplify and/or detect the presence of an AD nucleic acid, a probe or primer or set of such probes or primers or a combination thereof may include at least 1 polymorphism, or at least 2 polymorphisms, or at least 3 polymorphisms, or at least 4 polymorphisms associated with AD-related disease as shown in Tables A, B, C, D, and/or E, complementary sequences thereto, or polymorphisms that are in linkage disequilibrium with (genetically linked to) the polymorphisms in Tables A, B, C, D, and/or E (e.g. in the same haplotype block). To isolate, amplify and/or detect the presence of a nucleic acid associated with resistance to AD-related disease, a probe or primer or set of such probes or primers may include at least 1 allele, or at least 2 alleles, or at least 3 alleles, or at least 4 alleles associated with resistance to AD-related disease as shown in Tables A, B, C, D, and/or E, complementary sequences thereto, or alleles that are in linkage disequilibrium with (genetically linked to) the alleles in Tables A, B, C, D, and/or E (e.g. in the same haplotype pattern). To isolate, amplify and/or detect the presence of a nucleic acid associated with susceptibility to AD-related disease, a probe or primer or set thereof preferably includes at least 1 allele, or at least 2 alleles, or at least 3 alleles, or at least 4 alleles associated with susceptibility to AD-related disease as shown in Tables A, B, C, D, and/or E, complementary sequences thereto, or alleles that are in linkage disequilibrium with (genetically linked to) the alleles in Tables A, B, C, D, and/or E (e.g. in the same haplotype pattern).

In one embodiment, a probe or primer is at least about 70% identical to at least a portion of a nucleotide sequence (or complement thereof) that is being screened for the presence of an associated genomic region, preferably at least about 80% identical, more preferably at least about 90% identical, even more preferably about 95% identical, or even 100% identical. In any embodiment, a probe or primer may be optionally labeled with, for example, a radioactive, fluorescent, biotinylated or chemiluminescent label (e.g., radioisotope, fluorescent compound, enzyme, or enzyme co-factor.) Labeled nucleic acids are useful for detection of a hybridization complex and can be used as probes for diagnostic and screening assays.

Labeled probes can be used in cloning of full-length cDNA or genomic DNA by screening cDNA or genomic DNA libraries. Classical methods of constructing cDNA libraries are taught in Sambrook et al., supra. These methods provide for the production of cDNA from mRNA and the insertion of the cDNA into viral or other expression vectors. Typically, libraries of mRNA comprising poly(A) tails can be produced with poly(T) primers. Similarly, cDNA libraries can be produced using the nucleic acids herein as primers. Libraries of cDNA can be made either from selected tissues (e.g., normal or diseased tissue), or from tissues of a mammal treated with, for example, a pharmaceutical agent. Alternatively, many cDNA libraries are available commercially. In a preferred embodiment, the cDNA library is made from diseased or healthy human neuronal tissues or cells. In another preferred embodiment, members of the cDNA library are larger than a nucleic acid hybridization probe, and preferably contain the whole cDNA native sequence.

Genomic DNA can be isolated in a manner similar to the isolation of full-length cDNA. Briefly, the nucleic acids herein, or fragments, derivatives or complements thereof, can be used to probe a library of genomic DNA. Preferably, a genomic DNA library is obtained from neuronal tissue or cells but this is not essential. Such libraries can be in vectors suitable for carrying large segments of a genome, such as Pl or YAC, as described in detail in Sambrook et al., 9.4-9.30. In addition, genomic sequences can be isolated from human BAC libraries, which are commercially available from Research Genetics, Inc., Huntsville, Ala., USA, for example. As an alternative, full-length cDNA, genomic DNA, or any nucleic acid, fragment, derivative or complement thereof, can be obtained by synthesis. 2. Antisense and RNAi

Antisense nucleic acids, or mimetics thereof that are complementary, in whole or in part, to one or more AD nucleic acids are provided. Antisense nucleic acids can be used in diagnostics, prognostics, theranostics and/or treatment of AD-related disease. Antisense nucleic acids hybridize under high stringency conditions to target nucleic acids (e.g., associated genomic regions or RNA derivatives thereof such as mRNA). An antisense nucleic acid can bind RNA to form a duplex or a double-stranded DNA to form a triplex, which may be assayed. Preferably, hybridization of an antisense nucleic acid can act directly to block the translation of mRNA associated with susceptibility to AD-related disease by hybridizing to targeted mRNA and preventing protein translation. Absolute complementarity, although preferred, is not required. Antisense nucleic acids complementary to non-coding target nucleic acids associated with susceptibility to AD-related disease may also be used to inhibit translation of endogenous mRNA associated with susceptibility to AD-related disease by hybridizing to DNA regions involved in the transcription of the mRNA (e.g., regulatory regions, promoters, enhancers, etc.) While antisense nucleic acids complementary to a coding region sequence could be used, those complementary to the transcribed, untranslated region are most preferred. Antisense nucleic acids are preferably at least 10 nucleotides in length, more preferably at least 20 nucleotides, even more preferably at least 40 nucleotides in length, or more preferably at least 80 nucleotides in length. An antisense nucleic acid can be labeled for convenient detection, such as by using a radioisotope, fluorescent compound, enzyme or an enzyme co-factor. Regardless of the choice of target sequence, it is preferred that in vitro studies be first performed to quantify the ability of the antisense nucleic acid to inhibit mRNA expression. It is preferred that these in vitro studies utilize controls that distinguish between antisense inhibition and nonspecific biological effects of nucleic acids in a sample. Additionally, it is envisioned that results obtained using the antisense nucleic acid be compared with those obtained using a control nucleic acid. A control nucleic acid is preferably of approximately the same length as the test antisense nucleic acid and differs from the antisense nucleic acid sequence no more than is necessary to prevent specific hybridization to the target sequence.

The antisense nucleic acids herein can be modified at the base moiety, sugar moiety or phosphate backbone to improve stability of the molecule. Furthermore, the antisense nucleic acids may be hybridized or conjugated to another molecule (e.g., a peptide, hybridization triggered cross-linking agent, cleavage agent or transport agent) for targeting in a host cell or to facilitate the transport across the cell membrane (see, e.g., Letsinger et al. (1989) Proc. Natl. Acad. Sci. USA 86:6553-6556; Lemaitre et al., (1987), Proc. Natl. Acad. Sci. USA 84:648-652); for blood-brain barrier (see, e.g., PCT Publication No. W089/10134); to facilitate the hybridization-triggered cleavage agents (see, e.g., Krol et al. (1988) BioTechniques 6:958-976) or intercalating agents (see, e.g., Zon, (1988), Pharm. Res. 5:539- 549).

The antisense nucleic acids may comprise at least one modified base moiety which is selected from the group including but not limited to 5-fluorouracil, 5-bromouracil, 5- chlorouracil, 5-iodouracil, hypoxanthine, xanthine, 4-acetylcytosine, 5- (carboxyhydroxylmethyl)uracil, 5-carboxymethylaminomethyl-2-thiouridine, 5- carboxymethylaminomethyluracil, dihydrouracil, beta-D-galactosylqueosine, inosine, N6- isopentenyladenine, 1-methylguanine, 1-methylinosine, 2,2-dimethylguanine, 2- methyladenine, 2-methylguanine, 3-methylcytosine, 5-methylcytosine, N6-adenine, 7- methylguanine, 5-methylaminomethyluracil, 5-methoxyaminomethyl-2-thiouracil, beta-D- mannosylqueosine, 5'-methoxycarboxymethyluracil, 5-methoxyuracil, 2-methylthio-N6- isopentenyladenine, uracil-5-oxyacetic acid (v), wybutoxosine, pseudouracil, queosine, 2- thiocytosine, 5-methyl-2-thiouracil, 2-thiouracil, 4-thiouracil, 5-methyluracil, uracil-5- oxyacetic acid methylester, uracil-5-oxyacetic acid (v), 5-methyl-2-thiouracil, 3-(3-amino-3- N-2-carboxypropyl)uracil, (acp3)w, and 2,6-diaminopurine.

The antisense nucleic acid may also comprise at least one modified sugar moiety selected from the group including but not limited to arabinose, 2-fluoroarabinose, xylulose, and hexose. In yet another embodiment, the antisense nucleic acid comprises at least one modified phosphate backbone selected from the group consisting of a phosphorothioate, a phosphorodithioate, a phosphoramidothioate, a phosphoramidate, a phosphordiamidate, a methylphosphonate, an alkyl phosphotriester, and a formacetal or analog thereof. In yet another embodiment, the antisense nucleic acid is an α-anomeric oligonucleotide. An α-anomeric oligonucleotide forms specific double-stranded hybrids with complementary RNA in which, contrary to the usual β-units, the strands run parallel to each other (Gautier, et al, (1987) Nucl. Acids Res. 15:6625-6641). The oligonucleotide is a 2'-O- methylribonucleotide (Inoue, et al., (1987) Nucl. Acids Res. 15:6131-6148), or a chimeric RNA-DNA analogue (Inoue, et al., (1987) FEBS Lett. 215:327-330). Antisense nucleic acids (as well as other nucleic acids) herein may be synthesized by standard methods known in the art, e.g., by use of an automated DNA synthesizer (such as are commercially available from Biosearch, Applied Biosystems, etc.). As examples, phosphorothioate oligonucleotides may be synthesized by the method of Stein, et al. (1988) Nucl. Acids Res. 16:3209, and methylphosphonate oligonucleotides can be prepared by use of controlled pore glass polymer supports Sarin, et al., (1988) Proc. Natl. Acad. Sci. USA 85:7448-7451, etc. Alternately, an antisense nucleic acid can be produced biologically by placing a target nucleic acid in an expression vector in an antisense orientation or by using reverse transcriptase along with other reagents to construct the complementary DNA stand. Antisense nucleic acids should be delivered to cells that express the target nucleic acid in vivo A number of methods have been developed for delivering antisense DNA or RNA to cells, e g , antisense molecules can be injected directly into the tissue site, or modified antisense molecules, designed to target the desired cells (e g , antisense linked to peptides or antibodies which specifically bind receptors or antigens expressed on the target cell surface) can be administered systemically

A preferred approach to achieve intracellular concentrations of an antisense molecule sufficient to suppress translation of endogenous mRNAs utilizes a recombinant DNA construct in which the antisense oligonucleotide is placed under the control of a strong promoter (e g , pol III or pol II) The use of such a construct to transfect target cells in a patient will result in the transcription of sufficient amounts of single stranded KNAs which will form complementary base pairs with the endogenous sequence transcripts and thereby prevent translation of the mRNA sequence For example, a vector can be introduced e g , such that it is taken up by a cell and directs the transcription of an antisense RNA Such a vector can remain episomal or become chromosomally integrated, as long as it can be transcribed to produce the desired antisense RNA Such vectors can be constructed by recombinant DNA technology methods standard m the art Vectors can be plasmid, viral, or others known in the art, used for replication and expression m mammalian cells Expression of the sequence encoding the antisense RNA can be by any promoter known in the art to act in mammalian, preferably human cells Such promoters can be inducible or constitutive Such promoters include but are not limited to the SV40 early promoter region (Bernoist and Chambon, (1981) Nature 290 304-310), the promoter contained in the 3'-long terminal repeat of Rous sarcoma virus (Yamamoto, et al , (1980) Cell 22 787-797), the herpes thymidine kinase promoter (Wagner, et al , (198 l) Proc Natl Acad Sci USA 78 1441-1445), and the regulatory sequences of the metallothionein gene (Bnnster, et al , (1982) Nature 296 39-42) Any type of plasmid, cosmid, YAC or viral vector can be used to prepare the recombinant

DNA construct that can be introduced directly into the tissue site Alternatively, viral vectors can be used that selectively mfect the desired tissue, in which case administration may be accomplished by another route (e g , systemically)

In any of the embodiments herein, it may be necessary to compare the nucleotide sequence of the nucleic acid obtained, isolated, amplified, or cloned with that of a control

The percent identity of two nucleotide sequences can be determined, for example, by aligning the sequences for optimal comparison purposes The nucleotides at corresponding positions are compared and the percent identity between the two sequences is a function of the number of identical positions shared by the sequences (e g , percent identity = [(the number of identical positions/total number of positions) x 100]. In some embodiments, the length of a sequence aligned for comparison purposes is at least 30%, preferably at least 40%, more preferably at least 60%, and even more preferably at least 70%, 80%, or 90% of the length of the reference sequence or a full sequence gene. An actual comparison of two nucleic acid sequences can be accomplished by well-known methods, for example, using a mathematical algorithm. In one example, such a mathematical algorithm is described in Karlin et al., (1993) Proc. Natl. Acad. Sci. USA, 90:5873-5877. In another example, such mathematical algorithm is the algorithm of Myers and Miller, (1989) CABIOS. Additional algorithms for sequence analysis are known in the art and include ADVANCE and ADAM as described in Torellis and Robotti (1994) Comput. Appl. Biosci., 10:3-5 and FASTA described in Pearson and Lipman (1988) Proc. Natl. Acad. Sci. USA, 85:2444-8.

RNAi, or "RNA interference" is a technique in which exogenous, double-stranded RNA complementary to a known target mRNA are introduced into a cell to cause the degradation of the target mRNA, thereby reducing or silencing gene expression. This method of gene regulation has been demonstrated in Drosophila, Coenorhabditis elegans, plants, and in mammalian cell cultures. In mammalian cells, siRNAs ("small-interfering RNAs" that are double-stranded) are transfected into cells. siRNAs can be created using a phage enzyme known as "DICER" and a multi-protein siRNA complex termed "RISC" (RNA induced silencing complex). Briefly, duplexes of short (-19 nucleotides in length) RNAs with symmetric 2-nucleotide 3 '-overhangs (siRNAs) are introduced into a cell where they associate with specific proteins in a ribonucleoprotein complex, which scans the mRNA in the cell and degrades the mRNA target that is homologous to the siRNA, thereby preventing translation of the mRNA message and, therefore, synthesis of the protein encoded therein. For a review of RNAi techniques, see, e.g., Huppi, et al. (2005) "Defining and Assaying RNAi in Mammalian Cells", Molecular Cell 17(l):l-10; Grimm.et al. (2005) "Adeno- associated virus vectors for short hairpin RNA expression", Methods Enzymol 392:381-405; Bantounas, et al. (2004) "RNA interference and the use of small interfering RNA to study gene function in mammalian systems", J Molec Endocrin 33:545-557; Gene, et al. (2004) "RNA interference in neuroscience", Brain Res MoI Brain Res 132(2):260-270; and Campbell, et al. (2005) "RNA interference: past, present and future", Curr Issues MoI Biol 7(l):l-6. 3. Ribozymes, Knock-Outs and Triple Helices

Ribozyme molecules designed to catalytically cleave target mRNA transcripts can also be used to prevent translation of such mRNA. See, e.g., WO 90/11364; Sarver, et al., (1990) Science 247: 1222-1225.

Ribozymes are enzymatic RNA molecules capable of catalyzing the specific cleavage of RNA. See Rossi, (1994) Current Biology 4:469-471. The mechanism of ribozyme action involves sequence specific hybridization of the ribozyme to complementary target RNA, followed by an endonucleolytic cleavage event. The composition of ribozyme molecules must have one or more sequences complementary to the target mRNA and must include the well known catalytic sequence responsible for mRNA cleavage. See, e.g., U.S. Pat. No. 5,093,246.

While ribozymes that cleave mRNA at site-specific recognition sequences can be used to destroy target mRNAs, the use of hammerhead ribozymes is preferred. Hammerhead ribozymes cleave mRNAs at locations dictated by flanking regions which form complementary base pairs with the target mRNA. The sole requirement is that the target mRNA have the following sequence of two bases: 5'-UG-3 ' . The construction and production of hammerhead ribozymes are well known in the art and are described in Myers, "Molecular Biology and Biotechnology: A Comprehensive Desk Reference," (VCH

Publishers, New York, 1995) page 833; and in Haseloff and Gerlach, (1988), Nature 334:585- 591.

Preferably a ribozyme is engineered so that the cleavage recognition site is located near the 5-end of the target mRNA, i.e., to increase efficiency and minimize the intracellular accumulation of non-functional mRNA transcripts.

The ribozymes herein may further include RNA endoribonucleases, also known as "Cech-type ribozymes," such as the one which occurs naturally in Tetrahymena thermophila (known as the IVS, or L- 19 IVS RNA) and which has been extensively described in Zaug, et al., (1984) Science 224:574-578; Zaug and Cech, (1986) Science 231:470-475; Zaug, et al., (1986) Nature 324:429-433; PCT Publication No. WO 88/04300; Been and Cech, (1986) Cell 47:207-216.

As in the antisense approach, ribozymes can be composed of modified nucleic acids (e.g., for improved stability, targeting, etc.) and are preferably delivered to cells that express the target gene in vivo. A preferred method of delivery involves using a DNA construct encoding the ribozyme under the control of a strong constitutive promoter (e.g., pol III or pol II), so that transfected cells will produce sufficient quantities of the ribozyme to destroy endogenous target mRNA and inhibit translation. Because ribozymes, unlike antisense molecules, are catalytic, a lower intracellular concentration is required for efficiency.

Endogenous target gene expression can also be reduced by inactivating or "knocking out" the target nucleic acid (e.g., coding regions or regulatory regions of the target gene) using targeted homologous recombination. See Smithies, et al., (1985) Nature 317:230-234; Thomas and Capecchi, (1987) Cell 51:503-512; Thompson, et al., (1989) Cell 5:313-321. For example, a non-functional nucleic acid (or a completely unrelated DNA sequence) flanked by DNA homologous to the endogenous target nucleic acid can be used, with or without a selectable marker and/or a negative selectable marker, to transfect cells which express the target gene in vivo. Insertion of the DNA construct, via targeted homologous recombination, results in inactivation of the target gene. Such approaches can be used in humans provided the recombinant DNA constructs are directly administered or targeted to the required site in vivo using appropriate viral vectors.

Alternatively, endogenous expression of a target gene can be reduced by targeting deoxyribonucleotide sequences complementary to the regulatory region of the target gene (i.e., the target gene promoter and/or enhancers) to form triple helical structures which prevent transcription of the target gene in target cells in the body. See generally, Helene, (1991), Anticancer Drug Des., 6(6):569-584; Helene, et al., (1992), Ann. N. Y. Acad. Sci., 60:27-36; and Maher, (1992), Bioassays 14(12):807-815.

Nucleic acids to be used in triple helix formation for the inhibition of transcription should be single-stranded and composed of deoxyribonucleotides. The base composition of these oligonucleotides must be designed to promote triple helix formation via Hoogsteen base pairing rules, which generally require sizable stretches of either purines or pyrimidines to be present on one strand of a duplex. Nucleic acids may be pyrimidine-based, which will result in TAT and CGC+ triplets across the three associated strands of the resulting triple helix. The pyrimidine-rich molecules provide base complementarity to a purine-rich region of a single strand of the duplex in a parallel orientation to that strand. In addition, nucleic acid molecules may be chosen which are purine-rich, for example, contain a stretch of G residues. These molecules will form a triple helix with a DNA duplex that is rich in GC pairs, in which the majority of the purine residues are located on a single strand of the targeted duplex, resulting in GGC triplets across the three strands in the triplex.

Alternatively, the potential sequences that can be targeted for triple helix formation may be increased by creating a so-called "switchback" nucleic acid. Switchback nucleic acids are synthesized in an alternating 5'-3', 3'-5' manner, such that they base pair with first one strand of a duplex and then the other, eliminating the necessity for a sizable stretch of either purines or pyrimidines to be present on one strand of a duplex.

In instances wherein the antisense, ribozyme, "knock-out," and/or triple helix molecules described herein are utilized to inhibit gene expression (e.g., expression of nucleic acids associated with susceptibility to AD-related disease), it is possible that the technique may so efficiently reduce or inhibit the transcription (triple helix; knock-out) and/or translation (antisense, ribozyme) of mRNA that it may cause severe negative side effects. In such cases, to ensure that substantially normal levels of target gene products or desired gene products are maintained, nucleic acids which encode and polypeptides exhibiting a desired target gene activity (e.g., polypeptides associated with resistance to AD-related disease) may, be introduced into cells via gene therapy methods. The desired gene product should not contain sequences susceptible to antisense, ribozyme or triple helix treatments that are being utilized.

The antisense, ribozyme and triple helix molecules herein may be prepared by any method known in the art for the synthesis of DNA and RNA molecules. 4. Expression Vectors and Vectors

In certain embodiments, the nucleic acids herein are used to over-express polypeptides associated with resistance to AD-related disease. In another embodiment, the nucleic acids herein are used to underexpress polypeptides associated with susceptibility to AD-related disease. To overexpress a polypeptide, for example, a nucleic acid encoding the polypeptide of interest can be ligated to a regulatory sequence that can drive the expression of the polypeptide in the animal cell type of interest at a level that is higher than expression in the absence of such a construct. Such regulatory regions are well known. In another example, a non-coding nucleic acid (e.g., an intron or a regulatory nucleic acid) may be introduced to increase the production of a polypeptide of interest. To underexpress an endogenous polypeptide, a nucleic acid encoding a transcription factor or antisense RNA that down-regulates the polypeptide or a nucleic acid that produces, e.g., a variant or inactive polypeptide may be introduced into the genome of an animal such that the endogenous expression will be reduced or inactivated. In addition to, or in the alternative, a non-coding nucleic acid herein (e.g., an intron or a regulatory nucleic acid) may be introduced to override a native regulatory nucleic acid.

Any one or more of the nucleic acids herein can be inserted into a vector. A vector can be used, for example, to transfer nucleic acids or to express the inserted nucleic acids. In one embodiment, nucleic acids comprising an exon associated gene region of Iu, pvrl2, tomm40, apoE, apoCl , apoC2, apoC4, clptml or a homolog or fragment thereof can be inserted into an expression vector to express a partial or complete AD gene product. In another embodiments an exon associated gene region of a2bpl, ahsg, apoE, app, c9orf52, cacnalc, ckm, ctnnd2, cugbpl, dkfzp566kl924, farsl, fgl2, flj 14442, flj36760, flj38736, kiaal486, kiaal862, Iaptm4a, Inx2, Iocl47468, Iocl66522, Ioc283867, Ioc387711, Ioc388110, Ioc401237, lφlB, mata3, mgc3971, mrlc2, nce2, pcbp3, pdel IA, pfkfb2, ppplrl2b, psenl, pvrl2, secl3Ll, sox5, tgds, tomm40, ttll2 or homologs or fragments thereof can be inserted into an expression vector to express a partial or complete AD gene product.

An exonic associated genomic region can be in the coding region or outside the coding region. Expression vectors may be constructed using methods known in the art. Such methods include in vitro recombinant DNA techniques, synthetic techniques, in vivo genetic recombination, and other techniques described in Sambrook, J. et al. "Molecular Cloning, A Laboratory Manual," (Cold Spring Harbor Press, Plainview, N. Y. 1989), and Ausubel, F.M. et al. "Current Protocols in Molecular Biology", (John Wiley & Sons, New York, N. Y., 1989). A vector may also comprise one or more regulatory elements that direct the expression of a coding sequence in a host cell. Regulatory elements include but are not limited to inducible and non-inducible promoters, enhancers, operators, and other elements that drive and regulate expression.

There are numerous types of expression vectors. One type of expression vector is a plasmid, which refers to a circular double stranded DNA molecule into which additional DNA segments can be ligated. Another type of vector is a viral vector, wherein additional DNA segments can be ligated into a viral genome. Viral vectors include replication defective retroviruses, adenoviruses and adeno associated viruses. Certain vectors are capable of autonomous replication in a host cell into which they are introduced (e.g., bacterial vectors having a bacterial origin of replication and episomal mammalian vectors). Other vectors (e.g., non-episomal mammalian vectors) are integrated into the genome of a host cell upon introduction into the host cell, and thereby are replicated along with the host genome. Moreover, certain vectors, e.g., expression vectors, are capable of directing the expression of genes to which they are operably linked. A preferable expression vector is a plasmid, an artificial chromosome, a cosmid or a viral vector.

The expression vectors herein can include one or more regulatory sequences, selected on the basis of the host cells to be used and the level of expression desired. The regulatory sequences can be operably linked to the nucleic acid sequence to be expressed. The term operably linked refers to a nucleic acid of interest that is linked to one or more regulatory sequences in a manner that allows for the expression of the nucleic acid of interest. The term regulatory sequence includes promoters, enhancers and other expression control elements (e.g., polyadenylation signals). Such regulatory sequences are described, for example, in Goeddel, "Gene Expression Technology Methods in Enzymology" (1990) 185, Academic Press, San Diego, CA Regulatory sequences include those that direct constitutive expression of a nucleotide sequence in many types of host cells and those that direct expression of the nucleotide sequence only in certain host cells (e g , tissue-specific regulatory sequences)

In another embodiment, a coding region of an associated genomic region can be inserted into an expression vector with or without a non-coding region of interest The difference in expression or activity between a vector comprising both the non-coding and coding sequence can be detected using methods known in the art The vectors herein can be inserted into a host cell The term "host cell" refers not only to a particular subject cell but also to the progeny or potential progeny of such a cell Because certain modifications may occur in succeeding generations due to either mutations or environmental influences, such progeny may not, in fact, be identical to cells, but are still included within the scope of the term as used herein Vectors can be introduced into prokaryotic or eukaryotic cells via conventional transformation or transfection techniques For example, expression systems in bacteria include those described in Chang et al , (1978) Nature 275 615, and Siebenhst et al , (1980) Cell 20 269, expression systems in yeast include those described in Kelly and Hynes, EMBO J (1985) 4 475-479, expression systems in insect cells mclude those described in Maeda et al , (1985) Nature 315 592-594 and expression in mammalian cells include those described, for example, in Dijkema et al , (1985) EMBO J 4 761 Vector constructs can comprise either sense or antisense sequences, or both

As used herein, the terms transformation and transfection refer to a variety of art- recognized techniques for introducing a foreign nucleic acid molecule (e g , DNA) into a host cell, including calcium phosphate or calcium chloride co-precipitation, DEAF-dextran- mediated transfection, hpofection, or electroporation Suitable methods for transforming or transfectmg host cells can be found in Sambrook, et al and other laboratory manuals For stable transfection of mammalian cells, it is known that, depending upon the expression vector and transfection technique used, only a small fraction of cells may integrate the foreign DNA into their genome To identify and select these integrants, a gene that encodes a selectable marker is generally introduced into the host cells along with the gene of interest Preferred selectable markers include those that confer resistance to drugs Nucleic acid molecules encoding a selectable marker can be introduced into a host cell on the same vector as the nucleic acids or can be introduced on a separate vector Cells stably transfected with the introduced nucleic acid molecule can be identified by drug selection (e.g., cells that have incorporated the selectable marker gene will survive, while the other cells die).

A variety of host-expression vector systems may be utilized to express the AD coding nucleic acids of the invention. Such host expression systems represent not only the vectors by which the coding sequences may be expressed and their encoded RNAs or polypeptides purified, but also represent the cells containing these vectors. These include, but are not limited to bacteria (e.g., E. coli, B. subtilis) transformed with recombinant bacteriophage DNA, plasmid DNA or cosmid DNA expression vectors; yeast (e.g., Saccharomyces, Pichia) transformed with recombinant yeast expression vectors, insect cell systems transformed with recombinant viral expression vectors (e.g., baculovirus), plant cell systems transformed with recombinant viral expression vectors (e.g., cauliflower mosaic virus, tobacco mosaic virus) or transformed with recombinant plasmid expression vectors (e.g., Ti plasmid), and mammalian cell systems (e.g., COS, CHO, BHK, 293, 3T3 cell lines) transformed with recombinant expression constructs containing promoters derived from the genome of mammalian cells (e.g., metallothionein promoter) or from mammalian viruses (e.g., the adenovirus late promoter, vaccinia virus 7.5K promoter). Such vectors and host- expression vector systems are well known in the art and are further described in, e.g., Ruther et al. (1983), EMBO J. 2:1791; Inouye & Inouye (1985) Nucleic Acids Res. 13:3101-3109; Van Heeke & Schuster (1989) J. Biol. Chem. 264:5503-5509; Smith et al. (1983) J. Virol. 46:584; Smith, U.S. patent no. 4,215,051 ; Logan & Shenk (1984) Proc. Natl. Acad. Sci. USA 81 :3655-3659; Bittner et al. (1987) Methods in Enzymol. 153:516-544; Alam (1990) Anal. Biochem. 188:245-254; MacGregor & Caskey (1989) Nucl. Acids Res. 17:2365; and Norton & Corrin (1985) MoI. Cell. Biol. 5: 281.

In addition, a host cell may be chosen that modulates the expression of a vector- encoded nucleic acid sequence, or that modifies and processes an encoded RNA or polypeptide in a specific manner. Such modifications (e.g., glycosylation, phosphorylation) and processing (e.g., cleavage, folding) of polypeptides may be important for the function of the polypeptide. Different host cells have characteristic and specific mechanisms for the post-translational modification of polypeptides. Appropriate host cells or systems can be chosen to ensure that the correct modification and processing of an encoded protein. As such, in some situations, it may be desirable to express a eukaryotic gene in a eukaryotic cell where the gene will benefit from native folding and posttranslational modifications. Such mammalian host cells include, but are not limited to CHO, VERO, BHK, HeLa, COS, MDCK, 293, 3T3, W138, etc. Host cells can be used to produce polypeptides encoded by any of the nucleic acids herein. Suitable host cells and methods for producing polypeptides using such host cells are discussed in Goeddel, supra. For large scale protein production, a unicellular organism such as E. coli, baculoviras vectors, or cells of higher organisms such as vertebrates, particularly mammals, e.g. COS7 cells, may be useful. Host cells into which an expression vector has been introduced may be cultured in suitable medium such that the polypeptide is produced. The polypeptide herein may be isolated from the medium or from the host cell.

Host cells can also be used to produce nonhuman transgenic animals. For example, in one embodiment, a host cell is a fertilized oocyte or an embryonic stem cell into which a nucleic acid (e.g., an exogenous AD gene or a nucleic acid encoding a polypeptide herein) has been introduced. Such host cells can then be used to create non-human transgenic animals in which exogenous nucleotide sequences have been introduced into the genome or homologous recombinant animals in which endogenous nucleotide sequences have been altered. Such animals are useful for studying the function and/or activity of the nucleotide sequence and polypeptide encoded by the sequence and for identifying and/or evaluating modulators of their activity. As used herein, a "transgenic animal" is a non-human animal, preferably a mammal, more preferably a rodent such as a rat or mouse, in which one or more of the cells of the animal include a transgene. Other examples of transgenic animals include, for example, non-human primates, sheep, dogs, cows, goats, chickens and amphibians. A transgene is an exogenous DNA which is integrated into the genome of a cell from which a transgenic animal develops and which remains in the genome of the mature animal, thereby directing the expression of an encoded gene product in one or more cell types or tissues of the transgenic animal. As used herein, an homologous recombinant animal is a non-human animal, preferably a mammal, more preferably a mouse, in which an endogenous gene has been altered by homologous recombination between the endogenous gene and an exogenous DNA molecule introduced into a cell of the animal, e.g., an embryonic cell of the animal, prior to development of the animal.

Methods for generating transgenic animals via embryo manipulation and microinjection, particularly animals such as mice, are conventional in the art and are described, for example, in U.S. Patent Nos. 4,736,866, 4,870,009, 4,873,191 and in Hogan, "Manipulating the Mouse Embryo," (Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N. Y., 1986). Methods for constructing homologous recombination vectors and homologous recombinant animals are described further in Bradley (1991) Current Opinion in BioTechnology, 2:823-829. Clones of the non-human transgenic animals described herein can also be produced according to the methods described in Wilmut et al. (1997) Nature 385:810-813; and PCT Publication Nos. WO 97/07668 and WO 97/07669. Ill Polypeptides

AD polypeptides such as those encoded by or regulated by associated genomic regions comprising the variants, preferably, the susceptibility variants, identified in Tables A, B, C, D, and/or E. For example, AD polypeptides include those encoded by a TOMM40 exon encompassing a "POSITION" of 50087984 or any fragment, complement, derivative, homolog, or analog thereof. In certain embodiments, an AD polypeptide is one that is regulated by an intronic TOMM40 region that encompasses a "POSITION" of 50095698, 50095252, 50096902, and/or 50096271 or any fragment, complement, derivative, homolog, or analog thereof. In certain embodiments, an AD polypeptide herein is regulated by an intron of APOE that encompasses a "POSITION" of 50102284 or 50101007 or any fragment, complement, derivative, homolog, or analog thereof. In certain embodiments, an AD polypeptide is one that is encoded by an ND3 exon encompassing a "POSITION" of 10401 or any fragment, complement, derivative, homolog, or analog thereof. In certain embodiments, an AD polypeptide is one that is encoded by an ND4 exon encompassing a "POSITION" of 11915 or any fragment, complement, derivative, homolog, or analog thereof. In certain embodiments, an AD polypeptide herein is regulated by an intron of PVRL2 that encompasses a "BEST_POSITION" of 50053064. The AD polypeptides herein may be naturally occurring or recombinantly produced using methods known in the art.

An AD polypeptide can be associated with resistance or susceptibility to AD-related disease, including association with the age-of-onset of AD-related disease. In particular, an AD polypeptide associated with resistance to AD-related disease may increase the age-of- onset in an individual, and an AD polypeptide associated with susceptibility to AD-related disease may decrease the age-of-onset in an individual. A polypeptide associated with resistance to AD-related disease may be one that is expressed differently in individuals having a phenotype of resistance to AD-related disease as compared to individuals who do not have a phenotype of resistance to AD-related disease, or one that is regulated or encoded in whole or in part by a nucleic acid associated with resistance to AD-related disease. In one example, a polypeptide associated with AD-related disease can be recombinantly produced using an expression vector having a non-coding regulatory region associated with resistance to AD, operably linked to an AD polypeptide. The expression vector is introduced into a host cell under conditions appropriate for expression. The polypeptide can then be isolated from the host cell using standard protein purification techniques.

Similarly, a polypeptide associated with susceptibility to AD-related disease may be one that is expressed differently in individuals having a phenotype of susceptibility to AD- related disease as compared to individuals who do not have a phenotype of susceptibility to AD-related disease, or one that is regulated or encoded, in whole or in part, by nucleic acids associated with susceptibility to AD-related disease. For example, a polypeptide associated with AD-related disease can be recombinantly produced by introducing an expression vector with a coding nucleic acid associated with susceptibility to AD-related disease into a host cell. The host cell is maintained under conditions suitable for expression. The polypeptide is then isolated from the host cell.

In one embodiment, a polypeptide associated with resistance to AD-related disease can be produced by inserting a non-coding nucleic acid or nucleic acid outside coding region which is associated with resistance to AD-related disease, operably linked to an associated genomic region coding sequence, into a host cell under conditions appropriate for protein synthesis, and then purifying the polypeptide expressed by the host cell.

A similar method can be used to produce a polypeptide associated with susceptibility to AD-related disease. For example, a non-coding nucleic acid or nucleic acid outside coding region which is associated with susceptibility to AD-related disease, operably linked to an associated genomic region coding sequence, can be inserted into a host cell under conditions appropriate for protein synthesis. The resulting polypeptide associated with susceptibility to AD-related disease is then collected and purified.

In a preferred embodiment, a polypeptide associated with susceptibility to AD- related disease can be produced by inserting a vector comprising a coding nucleic acid associated with susceptibility or resistance to AD-related disease and then purifying the polypeptide expressed by the host cell.

In preferred embodiments, the polypeptides are purified. There are various degrees of purity. While a polypeptide can be purified to homogeneity, preparations in which a polypeptide is not purified to homogeneity are also useful where the polypeptide retains a desired function even in the presence of considerable amount of other components. In some embodiments, polypeptides are substantially free of cellular material which includes preparations of a polypeptide having less than about 30% (dry weight) other polypeptides (e.g., contaminating polypeptides), less than about 20% other polypeptides, less than about 10% other polypeptides, or less than about 5% other polypeptides.

When a polypeptide is recombinantly produced, it can also be substantially free of culture medium. In preferred embodiments, culture medium represents less than about 20% of the volume of the polypeptide preparation, preferably less than about 10% of the volume of the polypeptide preparation or more preferably less than about 5% of the volume of the polypeptide preparation. Polypeptides that are substantially free of chemical precursors or other chemicals generally include those that are separated from chemicals that are involved in its synthesis. In one embodiment, the polypeptides are substantially free of chemical precursors or other chemicals such that a preparation of the polypeptides has less than about 30% (dry weight) chemical precursors or other chemicals, preferably less than about 20% chemical precursors or other chemicals, more preferably less than about 10% chemical precursors or other chemicals or more preferably less than about 5% chemical precursors or other chemicals.

As used herein, two polypeptides are substantially homologous when their amino acid sequences are at least about 45% homologous, or preferably at least about 75% homologous, or more preferably at least about 85% homologous, or even more preferably greater than about 95% homologous. To determine the percent homology of two polypeptides, the amino acid sequences are aligned for optimal comparison purposes. The amino acid residues at corresponding positions are compared. The percent homology between two amino acid sequences is a function of the number of identical positions shared by the sequences (e.g., percent homology equals the number of identical positions/total number of positions times 100).

Some polypeptides (e.g., synonymous or conservative variants) may have a lower degree of sequence homology but are still able to perform one or more of the same functions. Conservative substitutions that can maintain the same function include replacements among aliphatic amino acids methionine, valine, leucine and isoleucine; interchange of the hydroxyl residues serine and threonine; exchange of acidic residues aspartic and glutamic acids; substitution between amide residues asparagine and glutamine, exchange between basic residues lysine and argmine, and replacements among aromatic residues phenylalanin, tyrosine and tryptophan. Alanine and glycine may also result in conservative substitutions. Other polypeptides that may not be able to perform one or more of the same functions may be variants containing one or more non-conservative amino acid substitutions or deletions, insertions, inversions or substitution of one or more amino acid residues. Amino acids that are essential for function of a polypeptide can be identified by various methods known in the art, such as site-directed mutagenesis or alanine-scanning mutagenesis. See Cunningham et al., (1989) Science, 244:1081-1085. The latter procedure can introduce a single alanine mutation at every residue in the molecule. The resulting variants are then tested for biological activity in vitro or in vivo. Residues that are critical for polypeptide activity or inactivity are identified by comparing the two variants (with and without the alanine mutation). Polypeptide activity can also be determined by structural analysis such as crystallization, nuclear magnetic resonance or photoaffinity labeling. See Smith et al, (1992) J. MoI. Biol., 224:899-904; and de Vos et al. (1992) Science, 255:306-312. 1. Fusion Proteins

Any polypeptides herein can be made part of a fusion protein. The term "fusion protein" or "fusion polypeptide" refers to an AD polypeptide (a polypeptide associated with resistance or susceptibility to AD) operatively linked to a non-AD polypeptide or a heterologous polypeptide having an amino acid sequence not substantially homologous to an AD amino acid sequence. "Operatively linked" indicates that the polypeptide and the heterologous protein are fused, for example, the non-AD polypeptide can be fused to the N- terminus or C-terminus of the AD polypeptide. In a preferred embodiment, the fusion polypeptide does not affect the function of the AD polypeptide. Examples of fusion polypeptide that do not affect the function of a polypeptide include a GST- fusion polypeptides in which the AD polypeptide sequences are fused to the C-terminus of the GST sequences. Other types of fusion polypeptides include enzymatic fusion polypeptides, for example β-galactosidase fusions, yeast two-hybrid GAL fusions, poly-His fusions and Ig fusions. Fusion polypeptides, especially poly-His fusions, can facilitate the purification of recombinant polypeptide. In some host cells, such as mammalian cells, expression and secretion of an AD polypeptide can be increased using a heterologous signal sequence. Therefore, in a preferred embodiment, an AD polypeptide may be fused to a heterologous signal sequence at its N-terminus. In another embodiment, a fusion protein may comprise of an AD polypeptide and various portions of immunoglobulin constant regions such as the Fc portion. Fc portions are useful in therapy and diagnosis and may result in improved pharmacokinetic properties. Fc portions can also be used in high-throughput screening assays to identify binding molecules, agonists and antagonists. See, e.g., Bennett et al.; J. of Molec. Recog., (1995) 8:52-58 and Johanson et al., (1995) J. of Biol. Chem., 270,16:9459-9471. In a preferred embodiment, soluble fusion proteins comprise of an AD polypeptide and one or more of the constant regions of heavy or light chains of immunoglobulins (e.g. IgG, IgM, IgA, IgD, IgE).

A fusion protein can be produced by standard recombinant DNA techniques as described herein. For example, DNA fragments coding for the different polypeptide sequences are ligated together in accordance with conventional techniques. The fusion gene can be synthesized by conventional techniques such as automated DNA synthesizers. Alternatively, PCR amplification of nucleic acid fragments can be carried out using anchor primers which give rise to complementary overhangs between two consecutive nucleic acid fragments that can subsequently be annealed and reamplified to generate a chimeric nucleic acid sequence. Moreover, many expression vectors are commercially available that already encode a fusion moiety (e.g., a GST protein). A nucleic acid encoding a polypeptide herein can be cloned into such an expression vector such that the fusion moiety is linked in-frame to the polypeptide. 2. Antibodies

Any of the polypeptides herein, or fragments, derivatives, or complements thereof, can be used as an immunogen (e.g. epitope) to generate polypeptide-specific antibodies. Antibodies can be used to detect, isolate and inhibit the activity of one or more AD polypeptides. To generate AD antibodies, an AD polypeptide or a fragment thereof is used as an epitope. In preferred embodiments, an epitope is at least 6 amino acids, at least 9 amino acids, at least 20 amino acids, at least 40 amino acids, or at least 80 amino acids in length. The epitope or polypeptide fragment preferably comprises a domain, segment or motif that can be identified by analysis using well-known methods, for example, signal polypeptides, extracellular domains, transmembrane segments or loops, ligand binding regions, zinc finger domains, DNA binding domains, acylation sites, glycosylation sites or phosphorylation sites.

Examples of antibodies contemplated by the present invention include polyclonal, monoclonal, humanized, chimeric, single chain antibodies, antibody fragments such as Fab fragments, F(ab')2 fragments, fragments produced by Fab expression library, anti-idiotypic (anti-Id) antibodies and epitope-binding fragments of any of the above.

Polyclonal antibodies are prepared by immunizing a suitable subject (e.g., goats, rabbits, rats, mice or humans) with a desired antigen. The antibody titer in the immunized subject can be monitored over time using methods known in the art, such as by using an enzyme linked immunosorbent assay (ELISA). The antibodies can then be isolated from the subject (e.g., from blood) and further purified using techniques, such as protein A chromatography, to obtain the IgG fraction.

At an appropriate time after immunization, such as when the antibody titers are highest, antibody-producing cells can be obtained from the subject and used for the preparation of monoclonal antibodies. Monoclonal antibodies are populations of antibodies that contain only one species of an antigen-binding site and are capable of immunoreacting with only one particular epitope of AD polypeptides. A monoclonal antibody composition, therefore, typically displays a single binding affinity for a particular polypeptide with which it immunoreacts. There are numerous methods known in the art for producing monoclonal antibodies.

In one example, monoclonal antibodies can be obtained by fusing individual lymphocytes (typically splenocytes) from an immunized animal (typically a mouse or a rat) with cells derived from an immortal B lymphocyte tumor (typically a myeloma) to produce a hybridoma. The culture supematants of the resulting hybridoma cells are screened to identify a hybridoma producing a monoclonal antibody that specifically binds to a polypeptide of interest. Other techniques for producing hybridoma include the human B cell hybridoma technique described in Kozbor et al. (1983) Immunol. Today, 4:72; the EBV-hybridoma technique and the trioma techniques.

Alternatively, monoclonal antibodies can be identified and isolated by screening a combinatorial immunoglobulin library, such as an antibody phage display library. The library can be screened with one or more of the polypeptides herein. Identified members are then isolated using techniques known in the art. Kits for generating and screening phage display libraries are commercially available. See for example, the Pharmacia Recombinant Phage Antibody System, Catalog No. 27-9400-01, and the Stratagene SurjZAPTM Phage Display Kit, Catalog No. 240612. Other methods and reagents for generating and screening antibody display libraries are disclosed in PCT Publication No. WO 92/01047; PCT Publication No. WO 90/02809; Fuchs et al. (1991) Bio/Technology, 9:1370-1372; Hay et al. (1992) Hum. Antibod. Hybridomas, 3:81-85; Huse et al. (1989) Science 246:1275-1281; Griffith et al. (1993) EMBO J. 12:725-734. The monoclonal antibodies can be chimeric and humanized. Humanized monoclonal antibodies can be obtained using standard recombinant DNA techniques in which the variable region genes (e.g., of a rodent antibody), are cloned into a mammalian expression vector containing the appropriate human light change and heavy chain region genes. In this example, the resulting chimeric monoclonal antibodies has the antigen-binding capacity from the variable region of the rodent but is significantly less immunogenic because of the humanized light and heavy chain regions. See, e.g., Surender K. Vaswani, Ann. (1998) Allergy Asthma. Immunol. 81 :105-119.

Any of the antibodies can further be coupled to a substance (label) for detection of a polypeptide-antibody binding complex. Examples of labels include, enzymes, prosthetic groups, fluorescent materials, luminescent materials, bioluminescent materials, or radioactive materials Examples of suitable enzymes include, for example, horseradish peroxidase, alkaline phosphatase, β-galactosidase, or acetylcholinesterase Examples of suitable prosthetic group complexes include, for example, streptavidin/biotin and avidm/biotin, examples of suitable fluorescent materials include umbelliferone, fluorescein, fluorescein isothiocyanate, rhodamine, dichlorotπazinylamme fluorescein, dansyl chloride or phycoerythrin An example of a luminescent material is luminol Examples of bioluminescent materials include luciferase, luciferin and aequoπn Examples of suitable radioactive material include 1251, 1311, 35S or 3H The antibodies can be used to isolate one or more AD polypeptides using standard techniques such as affinity chromatography or immunoprecipitation The antibodies can also be used to detect the presence or absence of a particular polypeptide (e g , a polypeptide associated with resistance or susceptibility to AD-related disease) in a cell, cell lysate, cell supernatant, tissue sample or elsewhere Preferably, the antibodies can further be used to inhibit or suppress the activity of such polypeptides by specifically binding to the polypeptides

Some antibodies of the invention specifically bmd to a protein encoded by a gene selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, AP0C4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, ND4, PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155,

LOC727827, FARPl, RNF113B, EX0C2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932, without specifically binding to another That is, antibodies specifically bind to a variant form having an amino acid encoded by a codon including a resistance allele shown in Table E without binding to a variant form having an amino acid encoded by a codon including a susceptibility allele shown in Table E, (or vice versa) Such antibodies are useful, for example, in assays described below to detect variant forms of the above proteins IV Diagnostic And Prognostic Assays

The nucleic acids, polypeptides, antibodies and other compositions herein may be utilized as reagents (e g , in pre-packaged kits) for prognosis and diagnosis of susceptibility or resistance to AD-related disease, and in particular LOAD (late-onset Alzheimer's disease) The methods can be practiced on subjects known to have one or more symptoms of an AD- related disease, such as dementia, as part of a differential diagnosis or prognosis of other diseases The methods can also be practiced on subjects having a known susceptibility to an AD-related disease The polymorphic profile of such an individual can increase or decrease the assessment of susceptibility or predicted age-of-onset of the disease. For example, an individual having two siblings with Alzheimer's disease is known to be at increased susceptibility to the disease compared with the general population. A finding of additional factors favoring susceptibility increases the risk whereas finding factor favoring resistance decreases the risk. Additional methods for using the polymorphic loci (e.g., SNPs and DIPs) to determine an individual's risk of exhibiting or developing AD-related disease (e.g., a form of Alzheimer's disease) are provided, e.g., in U.S. Patent No. 7,127,355, and U.S. patent application no. 11/510,261, both of which are entitled "Methods for Genetic Analysis" and are incorporated herein by reference in their entireties for all purposes. The invention provides methods of determining the polymorphic profile of an individual at one or more of SNPs of the invention. The SNPs includes those shown in Table E, and those in linkage disequilibrium with them. Those in linkage disequilibrium with them usually occur in the same genes or within 10 or 40 kb of the same genes. SNPs in linkage disequilibrium with the SNPs in Table E can be determined by haplotype mapping. Haplotypes can be determined by fusing diploid cells from different species. The resulting cells are partially haploid, allowing determination of haplotypes on haploid chromosomes (see US 20030099964). Alternatively, SNPs in linkage disequilibrium with exemplified SNPs can be determined by similar association studies to those described in the examples below. The polymorphic profile means the polymorphic forms occupying the various polymorphic sites in an individual. In a diploid genome, two polymorphic forms, the same or different from each other, usually occupy each polymorphic site. Thus, the polymorphic profile at sites X and Y can be represented in the form X(xl, xl), and Y (y 1, y2), wherein xl, xl represents two copies of allele xl occupying site X and yl, y2 represent heterozygous alleles occupying site Y.

The polymorphic profile of an individual can be scored by comparison with the polymorphic forms associated with resistance or susceptibility to Alzheimer's disease occurring at each site as shown in Table E. The comparison can be performed on at least 1, 2, 5, 10, 25, 50, or all of the polymorphic sites, and optionally, others in linkage disequilibrium with them. The polymorphic sites can be analyzed in combination with other polymorphic sites. However, the total number of polymorphic sites analyzed is usually fewer than 10,000, 1000, 100, 50 or 25.

The polymorphic profile is preferably determined at one or more polymorphic sites in each of at least 2, 5, 10, 15 or 20 of the following genes: APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUROG3, C10ORF35, LOC729099, ND3, ND4, PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. In some methods, the polymorphic profile includes at least 2, 5, 10, 15 or 20 sites in each of the following genes: APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUROG3, C10ORF35, LOC729099, ND3, ND4, PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Some methods determine a polymorphic profile in at least one site not within the APOE gene. Some methods determine a polymorphic profile in at least one polymorphic site not within one or more of the APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, or APOE genes. Some methods determine a polymorphic profile in at least one polymorphic site that does not occur within 40 kb of one or more of APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, or APOE. The number of resistance or susceptibility alleles present in a particular individual can be combined additively or as ratio to provide an overall score for the individual's genetic propensity to Alzheimer's and related diseases (see U.S.S.N. 60/566,302, filed April 28, 2004; U.S.S.N. 60/590,534, filed July 22, 2004; U.S. Patent No. 7,127,355; PCT US05/07375 filed March 3, 2005; U.S.S.N. 60/995,564, filed September 27, 2007; and U.S.S.N. 11/510,261, filed August 25, 2006. Resistance alleles can be arbitrarily each scored as +1 and susceptibility alleles as -1 (or vice versa). For example, if an individual is typed at 100 polymorphic sites of the invention and is homozygous for resistance at all of them, he could be assigned a score of 100% genetic propensity to resistance to Alzheimer's disease or 0% propensity to susceptibility to Alzheimer's disease. The reverse applies if the individual is homozygous for all susceptibility alleles. More typically, an individual is homozygous for resistance alleles at some loci, homozygous for susceptibility alleles at some loci, and heterozygous for resistance/susceptibility alleles at other loci. Such an individual's genetic propensity for Alzheimer's disease can be scored by assigning all resistance alleles a score of +1 , and all susceptibility alleles a score of -1 (or vice versa) and combining the scores. For example, if an individual has 102 resistance alleles and 204 susceptibility alleles, the individual can be scored as having a 33% genetic propensity to resistance and 67% genetic propensity to susceptibility. Alternatively, homozygous resistance alleles can be assigned a score of +1, heterozygous alleles a score of zero and homozygous susceptibility alleles a score of -1. The relative numbers of resistance alleles and susceptibility alleles can also be expressed as a percentage Thus, an individual who is homozygous for resistance alleles at 30 polymorphic sites, homozygous for susceptibility alleles at 60 polymorphic sites, and heterozygous at the remaining 63 sites is assigned a genetic propensity of 33% for resistance As a further alternative, homozygosity for susceptibility can be scored as +2, heterozygosity, as +1 and homozygosity for resistance as 0, or homozygosity for resistance can be scored as +2, heterozygosity, as +1 and homozygosity for susceptibility as 0 One of ordinary skill in the art will understand that the methods of determining a scoring scheme are not limited to these few examples and encompass other scoπng methodologies that are variations on these examples The individual's score, and the nature of the polymorphic profile are useful in prognosis or diagnosis of an individual's susceptibility to AD-related disease, or the likelihood of developing an AD-related disease at an early age Optionally, a patient can be informed of susceptibility to an AD-related disease indicated by the genetic profile Presence of a high genetic propensity to AD-related disease or an early age-of-onset thereof can be treated as a warning to commence prophylactic or therapeutic treatment Presence of a high propensity to disease also indicates the utility of performing secondary testing, such as testing brain activity by a psychometric test, such as the mmi-mental exam, or taking a biopsy

Polymorphic profiling is useful, for example, in selecting agents to effect treatment or prophylaxis of AD-related disease in a given individual Individuals having similar polymorphic profiles are likely to respond to agents in a similar way Several drugs such as tacrine (Cognex), donepezil (Aricept), πvastigmine (Exelon), or galantamine (Remmyl), memantine (Namenda) (Forrest Laboratories), NSAIDs, statins, and Vitamin E have been reported to have some beneficial effect Several other drugs are in clinical trials, such as, antigonadotropm-leuprohde (Voyager Pharmaceuticals), lecozotan SR (Wyeth), rasagiline mesylate (Eisai), TTP448 (TransTech Pharma), Ketasyn (Accera), Atomoxetme (Eh Lilly), AAB-001, an antibody to Aβ (Elan/Wyeth)

Polymorphic profiling is also useful for stratifying individuals in clinical trials of agents being tested for capacity to treat AD-related disease or related conditions Such trials are performed on treated or control populations having similar or identical polymorphic profiles (see EP99965095 5) Use of genetically matched populations eliminates or reduces variation in treatment outcome due to genetic factors, leading to a more accurate assessment of the efficacy of a potential drug Computer-implemented algorithms can be used to identify more genetically homogenous subpopulations in which treatment or prophylaxis has a significant effect notwithstanding that the treatment or prophylaxis is ineffective in more heterogeneous larger populations. In such methods, data are provided for a first population with an AD-related disease treated with an agent, and a second population also with the disease but treated with a placebo. The polymorphic profile of individuals in the two populations is determined in at least one polymorphic site in or within 40 kb or preferably 10 kb of a gene selected from the group consisting of APOE, APOC 1 , PVRL2, TOMM40, CLPTMl, AP0C2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. Data are also provided as to whether each patient in the populations reaches a desired endpoint indicative of successful treatment or prophylaxis. Subpopulations of each of the first and second populations are then selected such that the individuals in the subpopulations have greater similarity of polymorphic profiles with each other than do the individuals in the original first and second populations. There are many criteria by which similarity can be assessed. For example, one criterion is to require that individuals in the subpopulations have at least one susceptibility allele at each of at least ten of the above genes. Another criterion is that individuals in the subpopulations have at least 75% susceptibility alleles for each of the polymorphic sites at which the polymorphic profile is determined. Regardless of the criteria used to assess similarity, the endpoint data of the subpopulations are compared to determine whether treatment or prophylaxis has achieved a statistically significant result in the subpopulations. As a result of computer implementation, billions of criteria for similarity can be analyzed to identify one or a few subpopulations showing statistical significance.

Polymorphic profiling is also useful for excluding individuals not having AD- related disease from clinical trials. For example, diagnosis of Alzheimer's disease is usually based on psychometric tests which are subject to identifying false positives. Including such individuals in the trial increases the size of the population needed to achieve a statistically significant result. Individuals not having Alzheimer's disease can be identified by determining the numbers of resistances and susceptibility alleles in a polymorphic profile as described above. For example, if a subject is genotyped at ten sites in ten genes of the invention associated with Alzheimer's disease, twenty alleles are determined in total. If over 50% and preferably over 60% or 75% percent of these are resistance genes, the individual is unlikely to have Alzheimer's disease and can be excluded from the trial regardless of the presence of symptoms that may appear to resemble Alzheimer's disease. Polymorphic profiles can also be used after the completion of a clinical trial to elucidated differences in response to a given treatment. For example, the set of polymorphisms can be used to stratify the enrolled patients into disease sub-types or classes. It is also possible to use the polymorphisms to identify subsets of patients with similar polymorphic profiles who have unusual (high or low) response to treatment or who do not respond at all (non-responders). In this way, information about the underlying genetic factors influencing response to treatment can be used in many aspects of the development of treatment (these range from the identification of new targets, through the design of new trials to product labeling and patient targeting). Additionally, the polymorphisms can be used to identify the genetic factors involved in adverse response to treatment (adverse events). For example, patients who show adverse response may have more similar polymorphic profiles than would be expected by chance. This allows the early identification and exclusion of such individuals from treatment. It also provides information that can be used to understand the biological causes of adverse events and to modify the treatment to avoid such outcomes. Polymorphic profiles can also be used for other purposes, including paternity testing and forensic analysis as described by US 6,525,185. In forensic analysis, the polymorphic profile from a sample at the scene of a crime is compared with that of a suspect. A match between the two is evidence that the suspect in fact committed the crime, whereas lack of a match excludes the suspect. The present polymorphic sites can be used in such methods, as can other polymorphic sites in the human genome.

Polymorphic profiles can be used in further association studies of traits related to Alzheimer's disease including the AD-related diseases described above.

Although polymorphic profiling can be done at the level of individual polymorphic sites as described above, a more sophisticated analysis can be performed by analyzing haplotype blocks containing SNPs of the invention and/or others in linkage disequilibrium with them (see US 20040220750). In some instances, the boundaries of a haplotype block can be approximated by the length of a gene in which a polymorphic site occurs plus ten kb of flanking genomic sequence at either end. Each haplotype block can be characterized by two or more haplotypes (i.e., combinations of polymeric forms). In some instances, a haplotype can be determined by detecting a single haplotype-determining polymorphic form within a haplotype block. In other instances, multiple polymorphic forms are determined within the block (see Patil et al., Science 2001 Nov 23;294(5547):1719-23). The haplotype at each of the haplotype blocks containing SNPs of the invention in an individual is a factor in determining resistance or susceptibility to an AD-related disease in an individual, and can be characterized as associating with resistance or susceptibility as can individual polymorphic forms. The number of haplotype blocks occupied by haplotypes associated with resistance and the number associated with susceptibility in a particular individual can be combined additively as for individual polymorphic forms to arrive at a percentage representing genetic propensity to resistance or susceptibility to an AD-related disease. The measure is more accurate than simply combining individual polymorphic forms because it gives the same weight to haplotype blocks containing multiple polymorphic sites as haplotype blocks within a single polymorphic site. The multiple polymorphic forms within the same block are associated with the same propensity for resistance or susceptibility to an AD-related disease, and should not be given the same weight as multiple polymorphic forms in different haplotype blocks, which indicate independent resistance or susceptibility to an AD-related disease.

The haplotype blocks used in polymorphic profiling preferably include one or more of the following genes: APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUROG3, C10ORF35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. In some methods, each haplotype includes a different one of the above genes. Some methods determine a polymorphic profile in at least one polymorphic site in at least one haplotype block of the invention that does not include APOE. Some methods determine a polymorphic profile in at least one polymorphic site in at least one haplotype block of the invention that does not include at least one of APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, APOC4, or APOE. Some methods determine a polymorphic profile in at least one polymorphic site that does not occur within 40 kb of at least one of APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, APOC4, or APOE.

The methods of the invention detect haplotypes in at least 1, 2, 5, 10, 25, 50 or all of the haplotype blocks of the invention, preferably selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EX0C2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. The haplotypes can be detected in combination with haplotypes at haplotype blocks other than those of the invention. However, the number of haplotype blocks is typically fewer than 1000 and often fewer than 100 or 50.

The invention also provides methods of expression profiling by determining levels of expression profiling of one or more genes of the invention i.e., APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUROG3, C10ORF35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, KNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. The methods preferably determine expression levels of at least 2, 5, 10, 15, 20 or all of the above genes. Some methods determine the expression levels of at least one gene other than APOE, APOCl, PVRL2, TOMM40, CLPTMl , APOC2, or APOC4. Optionally, expression levels of other genes beyond those associated with AD-related disease in the present application are also determined. The expression levels of one or more genes in discrete sample (e.g., from a particular individual or cell line) are referred to as an expression profile. Typically, the expression profile is compared with an expression profile of the same genes in a control sample. The control sample can be a negative control (e.g., an individual (or population of individuals) not having or susceptible to an AD-related disease) or a positive control (e.g., an individual (or population of individuals) having or susceptible to an AD-related disease). The controls can be contemporaneous or historical. Individual expression levels in both the test and control samples can be normalized before comparison, e.g., by reference to the levels of a house keeping genes to avoid differences unrelated to the disease. The relative similarity of the expression profile of a test individual to the negative and positive control expression profiles is a measure of the individual's resistance or susceptibility to an AD-related disease. For example, if an expression profile is determined for ten genes of the invention, and the expression levels in the test subject are more similar to the positive control than the negative control for nine of the gene, one can conclude that the test individual has or is susceptible to an AD-related disease. The analysis can be performed at a more sophisticated level by weighting expression level according to where they lie between negative and positive controls. For example, if there is a large difference between negative and positive controls, and an expression level of a particular gene in a test individual lies close to the positive control that expression level is accorded greater weight than an expression level in a gene in which there is a smaller difference in expression levels between negative and positive controls, and the expression level of the test individual lies only slightly above the midpoint of the negative and positive control expression levels.

A variety of methods may be used to prognosticate and diagnose susceptibility or resistance to AD-related disease. The following methods are provided as examples and not as limitations of means to diagnose AD-related disease. 1. Detection of AD Nucleic Acids

Detection of presence or increased level of one or more nucleic acids, or fragments, derivatives, variants or complements thereof, associated with resistance to AD-related disease is a prognostic and diagnostic for resistance to AD-related disease. On the other hand, detection of presence or increased level of one or more nucleic acids, or fragments, derivatives, variants or complements thereof, associated with susceptibility to AD-related disease is a prognostic and diagnostic for susceptibility to AD-related disease. Similarly, detection of the presence of a genetic variant (e.g., SNP) associated with AD-related disease may be used to diagnose the disease, while detection of a variant correlated with resistance may be indicative of a healthy state.

Detection of nucleic acids and genetic variations in an individual may be made using any method known in the art. Examples of such methods include, for instance, Southern or Northern analyses, in situ hybridizations analyses, single stranded conformational polymorphism analyses, polymerase chain reaction analyses and nucleic acid microarray analyses. Such analyses may reveal both quantitative and qualitative aspects of the expression pattern of AD polypeptides. In particular, such analyses may reveal expression patterns or polypeptides associated with resistance or susceptibility to AD-related disease. In one example, a diagnosis or prognosis is made using a test sample containing genomic DNA or RNA obtained from the individual to be tested. The individual can be an adult, child or fetus. The individual is preferably a human. The test sample can be from any source which contains genomic DNA or RNA including, e.g., blood, amniotic fluid, cerebrospinal fluid, skin, muscle, buccal or conjunctival mucosa, placenta, gastrointestinal tract or other organs. A test sample of DNA from fetal cells or tissue can be obtained by appropriate methods such as by amniocentesis or chorionic villus sampling. The test sample is subjected to one or more tests to identify the presence or absence of a nucleic acid of interest or a genetic variant of interest.

In one embodiment, Southern blot, northern blot or similar analyses methods are used to identify the presence or absence of a nucleic acid of interest or a genetic variant of interest using complementary nucleic acid probes associated with AD-related disease. The nucleic acid probes are preferably labeled before contacted with the sample.

In hybridization analysis, the sample is maintained under conditions sufficient to allow for specific hybridization of the nucleic acid probe to the target nucleic acid. In a preferred embodiment, the labeled nucleic acid probe and target nucleic acid specifically hybridize with no mismatches. Specific hybridization can be performed under stringent conditions disclosed herein and can be detected using standard methods. Hybridization is indicative of the presence or absence of a target nucleic acid. Specific hybridization to a nucleic acid or variant associated with resistance to AD-related disease is a diagnostic for resistance to AD-related disease. Specific hybridization to a nucleic acid or variant associated with susceptibility to AD-related disease is a diagnostic for susceptibility to AD- related disease. More than one probe can be used concurrently.

In a preferred embodiment, a nucleic acid probe is an allele-specific probe. See Saild, R. et al., (1986) Nature 324:163-166. Allele-specific probes can used to identify the presence or absence of one or more variants in a test sample of DNA obtained from an individual. A target nucleic acid is amplified using any method herein. Flanking sequences may also be amplified. In the case of Southern analysis, the amplified target nucleic acid is dot-blotted, using standard methods and the blot is then contacted with an allele specific nucleic acid probe. See Ausubel, F. et al., "Current Protocols in Molecular Biology" (eds. John Wiley & Sons). Detection of specific hybridization of an allele-specific probe to a target nucleic acid associated with resistance to AD-related disease is a diagnostic for resistance to AD-related disease. Detection of specific hybridization of an allele-specific probe to a target nucleic acid associated with susceptibility to AD-related disease is a diagnostic for susceptibility to AD-related disease.

Allele-specific probes are nucleic acids, mimetics, or a combination thereof, of approximately 10-50 base pairs or more preferably approximately 15-30 base pairs that specifically hybridize to one or more target nucleic acids. Target nucleic acids are any of the nucleic acids herein.

In one example, a target nucleic acid is a nucleic acid associated with resistance to AD. Nucleic acid probes that may be useful in identifying such target can be complementary to 1 or more, 2 or more, 3, or more, 4 or more, or 5 or more variants associated with resistance to AD-related disease. In another example, a target nucleic acid is a nucleic acid associated with susceptibility to AD-related disease. Nucleic acid probes that may be useful in identifying such target can be complementary to 1 or more, 2 or more, 3, or more, 4 or more, or 5 or more variants associated with susceptibility to AD-related disease. Such nucleic acid probes may be part of a set or in a kit (e.g., for use in Southern analysis or other techniques). Such nucleic acid probes can be allele-specific. Methods for preparing allele specific probes are known in the art.

One method for detecting nucleic acids associated with resistance or susceptibility to AD-related disease is northern analysis. Northern analysis can be used to identify gene expression patterns (e.g., levels of mRNA expression in different cell types or tissues, or during different developmental stages) of AD nucleic acids. See Ausubel, F. et al., "Current Protocols in Molecular Biology" (eds. John Wiley & Sons 1999). For northern analysis, a test sample of RNA is obtained from an individual by appropriate means. Specific hybridization of the test sample of RNA to a nucleic acid probe that is complementary to an RNA sequence associated with resistance to AD-related disease (e.g., encoding a polypeptide associated with resistance to AD-related disease) is a diagnostic or prognostic for resistance to AD-related disease. Specific hybridization of the test sample of RNA to a nucleic acid probe that is complementary to an RNA sequence associated with susceptibility to AD-related disease (e.g., encoding a polypeptide associated with susceptibility to AD-related disease) is a diagnostic or prognostic for susceptibility to AD-related disease. A nucleic acid probe is preferably labeled for northern blot analysis. A nucleic acid probe is preferably an allele- specific probe complementary to one or more of the variants (or polymorphisms) described in Tables A, B, C, D, and/or E, or may include kits or collections of probes with more than one of such probes.

Alternative diagnostic and prognostic methods employ amplification of target nucleic acids associated with resistance or susceptibility to AD-related disease, e.g., by PCR. This is especially useful for target nucleic acids present in very low quantities. In one embodiment, amplification of target nucleic acids associated with resistance to AD-related disease indicates their presence and is a prognostic and diagnostic of resistance to AD-related disease. In a related embodiment, amplification of target nucleic acids associated with susceptibility to AD-related disease indicates their presence and is a prognostic and diagnostic of susceptibility to AD-related disease.

In another embodiment, cDNA is obtained from test sample RNA nucleic acids by reverse transcription. Nucleic acid sequences within the cDNA may be used as templates for amplification reactions. Nucleic acids used as primers in the reverse transcription and amplification reaction steps can be chosen from any of the nucleic acids herein. For detection of amplified products, the nucleic acid amplification may be performed using labeled primers or labeled nucleotides. Alternatively, enough amplified product may be made such that the product may be visualized by standard ethidium bromide staining or by utilizing other suitable nucleic acid staining methods Alternatively, the amplified product may be labeled subsequent to the amplification reaction by conventional methods (e g , end-labeling)

The above-described methods for determining expression patterns of AD genes may also be performed on an isolated cell population of a particular cell type derived from a given tissue Additionally, in situ hybridization techniques may be utilized to provide information regarding which cells within a given tissue express an AD nucleic acid Such analyses may provide information regarding a specific biological function of an AD nucleic acid, and any genes or genomic regions in linkage equilibrium therewith Microarrays can also be utilized for diagnosis and prognosis of resistance or susceptibility to AD-related disease Microarrays comprise probes that are complementary to target nucleic acid sequences from an individual A microarray probe is preferably allele- specific In one embodiment, the microarray comprises a plurality of different probes, each coupled to a surface of a substrate in different known locations and each, capable of binding complementary strands See, e g , U S Pat No 5,143,854 and PCT Publication Nos WO

90/15070 and WO 92/10092 These microarrays can generally be produced using mechanical synthesis methods or light directed synthesis methods that incorporate a combination of photolithographic methods and solid phase oligonucleotide synthesis methods See Fodor et al , (1991) Science 251 767-777, and U S Pat No 5,424,186 Techniques for the mechanical synthesis of microarrays are described in, for example, U S Pat No 5,384,261

Once a microarray is prepared, one or more target nucleic acids are hybridized to the microarray before the microarray is scanned Typical hybridization and scanning procedures are described in PCT Publication Nos WO 92/10092 and WO 95/11995, and U S Pat No 5,424,186 Briefly, target nucleic acid sequences that include one or more previously identified variants or polymorphisms are amplified and labeled by well-known techniques, such as attachment of a fluorescent moiety or using labeled primers during amplification (e g PCR) Primers that are complementary to both strands of the target sequence (one pπmer complementary to one strand upstream and the other pnmer complementary to the other strand downstream from a variant or polymorphism) may be used to amplify the target region Asymmetric PCR techniques may be used An amplified target, preferably incorporating a label, is then hybridized with the microarray under appropriate conditions Upon completion of hybridization and washing of the microarray, the microarray is scanned to determine the position on the microarray to which the target sequence hybridizes The hybridization data obtained from the scan is typically in the form of fluorescence intensities as a function of location on the microarray.

Although primarily described in terms of a single detection block, such as for the detection of a single polymorphism, microarrays can include multiple detection blocks, and thus be capable of analyzing multiple specific polymorphisms. In an alternative arrangement, detection blocks may be grouped within a single microarray or in multiple separate microarrays so that varying optimal conditions may be used during the hybridization of the target to the microarray. For example, it may be desirable to provide for the detection of polymorphisms that fall within G-C rich stretches of a genomic sequence separately from those that fall in A-T rich segments for optimization of hybridization conditions. Additional description of use of nucleic acid microarrays for detection of polymorphisms can be found, for example, in U.S. Patent Nos. 5,858,659 and 5,837,832, the entire teachings of which are incorporated by reference herein.

Other methods to detect variant (or polymorphic) nucleic acids include, for example, direct manual sequencing (Church and Gilbert, (1988) Proc. Natl. Acad. Sci. USA 81 :1991- 1995; Sanger, F. et al. (1977) Proc. Natl. Acad. Sci. USA 74:5463-5467; and U.S. Pat. No. 5,288,644); automated fluorescent sequencing; single-stranded conformation polymorphism assays; clamped denaturing gel electrophoresis; denaturing gradient gel electrophoresis (Sheffield, V.C. et al. (1981) Proc. Natl. Acad. Sci. USA 86:232-236), mobility shift analysis (Orita, M. et al. (1989) Proc. Natl. Acad. Sci. USA 86:2766-2770), restriction enzyme analysis (Flavell et al. (1978) Cell 15:25; Geever, et al. (1981) Proc. Natl. Acad. Sci. USA 78:5081); heteroduplex analysis; Tm-shift genotyping (Germer et al. (1999) Genome Research 9:72-78); kinetic PCR (Germer et al. (2000) Genome Research 10:258-266); chemical mismatch cleavage (Cotton et al. (1985) Proc. Natl. Acad. Sci. USA 85:4397-4401); RNase protection assays (Myers, R.M. et al. (1985) Science 230:1242); and use of polypeptides which recognize nucleotide mismatches, such as E. coli mutS protein. 2. Detection of AD Polypeptides

Detecting the presence, level of expression, activity and location of AD polypeptides may be used as a diagnostic or prognostic for resistance or susceptibility to AD- related disease. Briefly, detection of the presence, level of expression or enhanced activity of polypeptides associated with resistance to AD-related disease is a diagnostic and prognostic for resistance to AD-related disease. Detection of the presence, level of expression or enhanced activity of polypeptides associated with susceptibility to AD-related disease is a diagnostic and prognostic for susceptibility to AD-related disease.

Proteins may be analyzed from any tissue or cell type, and in some specific embodiments neuronal tissues are used Analyses can be made in vivo or in vitro In a preferred embodiment a biopsy (or tissue sample) is obtained from brain tissue (e g , from the hippocampus, or the neocortex, or the parietal or temporal lobes) of an individual to be tested

Methods to detect and isolate polypeptides include, for example, enzymes linked immunosorbent assays (ELISAs), lmmunoprecipitations, immunofluorescence, lmmunoblotting, Western blotting, spectroscopy, coloπmetry, electrophoresis and isoelectric focusing See U S Pat No 4,376,110, see also Ausubel, F et al , "Current Protocols in Molecular Biology" (Eds John Wiley & Sons, chapter 10) Protein detection and isolation methods employed may also be those described in Harlow and Lane (Harlow, E and Lane, D , "Antibodies A Laboratory Manual," Cold Spring Harbor Laboratory Press, Cold Spnng Harbor, N Y , 1998)

In one embodiment, the presence, amount and location of polypeptides associated with resistance to AD-related disease can be determined using a probe or an antibody that specifically binds one or more polypeptides associated with resistance to AD-related disease In another embodiment, the presence, absence, amount or location of a polypeptide associated with susceptibility to AD-related disease can be determined using a probe or antibody that specifically bind one or more polypeptides associated with susceptibility to AD-related disease

Antibodies, such as those described herein may be used to determine the presence of a polypeptide associated with resistance or susceptibility to AD-related disease

In a preferred embodiment, a probe or antibody is labeled directly or indirectly Direct labeling involves coupling (physically linking) a detectable substance to an antibody or a probe Indirect labeling involves the reactivity of the probe with another reagent that is directly labeled Examples of indirect labeling include, for example, detection of a primary antibody using a fluorescently labeled secondary antibody and end labeling of a DNA probe with biotm such that it can be detected with fluorescently labeled streptavidm

A solid support may be utilized to immobilize either the antibody or probe or the sample (e g , AD polypeptide) In one example, a sample may be immobilized onto a solid support such as nitrocellulose, which is capable of immobilizing cells, cell particles, or soluble proteins The support may then be washed with suitable buffers followed by treatment with a detectably labeled antibody The amount of bound labeled antibody on the solid support may then be detected by conventional means Well known supports include glass, polystyrene, polypropylene, polyethylene, dextran, nylon, amylases, natural and modified celluloses, polyacrylamides, gabbros, and magnetite The antibodies herein can be linked to an enzyme and used in an enzyme immunoassay. See Voller, "The Enzyme Linked Immunosorbent Assay (ELISA)", Diagnostic Horizons 2:1-7 (Microbiological Associates Quarterly Publication, Walkersville, Md. 1978); Maggio, "Enzyme Immunoassay" (CRC Press, Boca Raton, FIa. 1980); Ishikawa, et al., "Enzyme Immunoassay" (Kgaku Shoin, Tokyo, 1981). The enzyme which is bound to the antibody will react with an appropriate substrate, preferably a chromogenic substrate, in such a manner as to produce a chemical moiety which can be detected, for example, by spectrophotometric, fluorimetric or by visual means. Enzymes that can be used to label the antibody include, but are not limited to, malate dehydrogenase, staphylococcal nuclease, delta-5-steroid isomerase, yeast alcohol dehydrogenase, alpha-glycerophosphate, dehydrogenase, triose phosphate isomerase, horseradish peroxidase, alkaline phosphatase, asparaginase, glucose oxidase, beta-galactosidase, ribonuclease, urease, catalase, glucose-6- phosphate dehydrogenase, glucoamylase and acetylcholinesterase. Detection can be accomplished by calorimetric methods which employ a chromogenic substrate for the enzyme. Detection can also be accomplished by visual comparison of the extent of enzymatic reaction of a substrate in comparison with similarly prepared standards.

Detection may also be accomplished using any of a variety of other immunoassays. For example, by radioactively labeling the antibodies or antibody fragments, it is possible to detect wild type or mutant peptides through the use of a radioimmunoassay. See Weintraub, B , "Principles of Radioimmunoassays, Seventh Training Course on Radioligand Assay

Techniques" (The Endocrine Society, March, 1986). The radioactive isotope can be detected by such means as the use of a gamma counter or a scintillation counter or by autoradiography.

It is also possible to label the antibody with a fluorescent compound. When the fluorescently labeled antibody is exposed to light of the proper wavelength, its presence can be detected my measuring emitted fluorescence. Among the most commonly used fluorescent labeling compounds are fluorescein isothiocyanate, rhodamine, phycoerythrin, phycocyanin, allophycocyanin, o-phthaldehyde and fluorescamine. The fluorescently labeled antibody can be coupled with light microscopic, flow cytometric or fluorimetric detection. In one example, antibodies, or fragments thereof, may be employed histological, as in immunofluorescence or immunoelectron microscopy, for in situ detection of a polypeptide associated with resistance or susceptibility to AD-related disease. In situ detection may be accomplished by removing a histological specimen from a patient, such as by biopsy. The specimen is then contacted with a labeled antibody described herein. The antibody or fragment is preferably contacted by overlaying the labeled antibody or fragment onto the sample. This procedure allows for the determination of the presence, absence, amount and location of a polypeptide of interest.

The antibody can also be detectably labeled using fluorescence emitting metals such as 152Eu, or others of the lanthanide series. These metals can be attached to the antibody using such metal chelating groups as diethylenetriaminepentacetic acid (DTPA) or ethylenediaminetetraacetic acid (EDTA).

The antibody also can be detectably labeled by coupling it to a chemiluminescent compound. The presence of the chemiluminescent-tagged antibody is then determined by detecting the presence of luminescence that arises during the course of a chemical reaction. Examples of particularly useful chemiluminescent labeling compounds are luminol, isoluminol, theromatic acridinium ester, imidazole, acridinium salt and oxalate ester.

Likewise, a bioluminescent compound may be used to label the antibodies herein. Bioluminescence is a type of chemiluminescence found in biological systems in which a catalytic protein increases the efficiency of the chemiluminescent reaction. The presence of a bioluminescent protein is determined by detecting the presence of luminescence. Preferred bioluminescent compounds for purposes of labeling antibodies are luciferin, luciferase and aequorin.

In one embodiment, the presence (or absence) of a polypeptide associated with AD- related disease in a sample (e.g., a cell, cell lysate, tissue, whether in vivo or in vitro) can be established by contacting the sample with an antibody and then detecting a binding complex. The presence of a polypeptide associated with resistance to AD-related disease is a diagnostic and prognostic of resistance to AD-related disease or more particularly LOAD. The presence of a polypeptide associated with susceptibility to AD-related disease is a prognosis and diagnosis of susceptibility to AD-related disease. In another embodiment, the level of expression or sequence of a polypeptide associated with AD-related disease in a test sample is compared with the level of expression or sequence of the same polypeptide in a control sample. A control sample may have a known level of expression of the polypeptide, and/or can be a sample from a healthy individual or from a different tissue or organ from the test individual. Alterations in the level of expression or sequence of an AD polypeptide may be indicative of susceptibility or resistance to AD-related disease. In one example, a test sample from an individual is assessed for a change in expression (e.g., level of transcription or translation) and/or sequence (e.g., splicing variants, polymorphisms) of a polypeptide associated with susceptibility to AD-related disease. Detection of an increased level of expression of a polypeptide associated with susceptibility to AD-related disease may be a prognosis or diagnosis of, for example, an onset of AD-related disease or an increased susceptibility to related disease. On the contrary, detection of a reduced level of a polypeptide associated with susceptibility to related disease may be indicative of, for example, a reduced susceptibility to AD-related disease or an effective treatment against AD- related disease (e.g., if the test sample is from an individual after treatment and the control sample is from the same individual before treatment). Detection of an increased level of a polypeptide associated with resistance to AD-related disease may be a prognosis or diagnosis of, for example, increased immunity to AD-related disease or an effective treatment regimen against AD-related disease. On the other hand, detection of a reduced level of a polypeptide associated with resistance to AD-related disease may be a prognosis or diagnosis of, for example, decreased immunity to AD-related disease or an ineffective treatment regimen against AD-related disease. Similarly, detection of an increase in compositions (including, e.g., peptides, derivatives, variants, splicing variants) associated with susceptibility to AD- related disease is a prognosis or diagnosis of an earlier onset or more severe symptoms of AD-related disease while detection of an increase in compositions associated with resistance to AD-related disease is a prognosis or diagnosis for immunity or reduced risk for developing AD-related disease.

Further, it may be useful to compare the level of expression of a reference AD polypeptide to the level of expression of an alternate or variant AD polypeptide in a cell or tissue that is heterozygous for a nonsynonymous polymorphism in a coding region. Such a cell or tissue may be expected to produce equivalent amounts of both the reference and alternate polypeptides encoded by the coding region. However, if measurement of the amounts of these two polypeptides indicates that one is produced at a statistically higher level than the other, then this is an indication that there is another regulatory mechanism at play. For example, it may be in indication that the coding region is exhibiting differential allelic expression, expressing one allele at a higher level than the other; that the RNA from one allele is being processed differently than the RNA for the other allele (e.g., via degradation, splicing, translation, etc.); or that the reference polypeptide is being processed differently than the alternate polypeptide (e.g., via degradation, post-translational modification, etc.) Kits useful in diagnosis and prognosis include reagents comprising, for example, instructions for use and analysis; means for collecting a tissue or cell sample; nucleic acid probes or primers (e.g., for amplification, reverse transcriptase and detection); labels (e.g., for nucleic acids or proteins); microarrays, gels, membranes or other detection apparati; restriction enzymes (e.g., for RFLP analysis); allele-specific probes; antisense nucleic acids; antibodies; and other protein binding probes, any of which may be labeled. V Screening Assays And Agents

The invention provides methods to identify agents potentially useful in diagnosis, prognosis, prophylaxis or treatment of an AD-related disease, including the likelihood of developing an AD-related disease at an early age. Agents are tested for their capacity to modulate expression or activity of a gene selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, AP0C4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, and ND4. Optionally, the group can further comprise other genes in Tables A, B, C, and/or D, e.g., PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932. In some methods, the gene is other than at least one of APOE, APOCl, PVRL2, TOMM40, CLPTMl, AP0C2, or APOC4. Expression assays are usually performed in cell culture, but can also be performed in animal models or in an in vitro transcription/translation system. The cell culture can be of primary cells, particularly, those known or suspected to have a role in AD-related disease, such as neurons or cells transfected with a gene of the above group. In the latter case, the coding portion of the gene is typically transfected with its naturally associated regulatory sequences, so as to permit expression of the gene in the transfected cell. However, the coding portion of the gene can also be operably linked to regulatory sequences from other (i.e., heterologous) genes. Optionally, the protein encoded by the gene is expressed fused to a tag or marker to facilitate its detection. The compound to be screened is introduced into the cell, usually in the form of a DNA molecule that can be expressed or directly as an RNA or protein. Expression of the gene can be detected either at the mRNA or protein level. Expression at the mRNA level can be detected by a hybridization assay, and at the protein level by an immunoassay. Detection of the protein level is facilitated by the presence of a tag. Similar screens can be performed in an animal, either natural or transgenic, or in vitro. Expression levels in the presence of an agent under test are compared with those in a control assay in the absence of compound, an increase or decrease in expression indicating that the compound modulates activity of the gene.

Assays to detect modulation of a protein encoded by a gene of the invention can also be performed. In some instances, a preliminary assay is performed to detect specific binding between an agent and a protein encoded by a gene of the invention. A binding assay can be performed between the agent and a purified protein, of if the protein is expressed extracellularly, between the agent and the protein expressed from a cell. Optionally, either the agent or protein can be immobilized before or during the assay. Such an assay reduces the pool of candidate agents for an activity assay. The nature of the activity assay depends on the activity of the gene. Agents that modulate expression or activity of the genes of the invention can then be tested in animal models for AD-related diseases. The animal models can be transgenic (as described below) or nontransgenic. Agents are tested in comparison with otherwise similar control assays except for the absence of the compound being tested. A reduction or inhibition of a sign or symptom of disease by an agent relative to a control indicates an agent has pharmacological activity potentially useful in treating the disease. The animal models used for such testing can be the novel animal models described below or conventional animal models of AD-related disease of which many are known. Such models include, for example, mice bearing a 717 (APP770 numbering) mutation of APP described by Games et al., supra, and mice bearing a 670/671 (APP770 numbering) Swedish mutation of APP such as described by McConlogue et al., US 5,612,486 and Hsiao et al., Science, 274, 99 (1996).

Agents that modulate expression or activity of the genes of the invention can also be screened in similar fashion in animal models of other diseases, particularly other AD-related diseases. For example, use of an animal model of Parkinson's disease is described by Hashimoto et al., Ann. N. Y. Acad. Sci. 991 :171-88 (2003), and of prion disease by Barrett et al., J. Virol. 77(15):8462-9 (2003), and of ALS by Kilveny et al. Nature Medicine 5, 347- 350 (1999).

Examples of agents include, but are not limited to: transcription factors, binding molecules, antisense nucleic acids, PNAs, mimetics, small or large organic or inorganic molecules, polypeptides (e.g., soluble peptides, or Ig-tailed fusion peptides), antibodies, as described above, (e.g., monoclonal, polyclonal, humanized, anti-idiotypic, chimeric or single chain antibodies, Fab, F(ab')2, Fab expression library fragments, and epitope-binding fragments thereof), fusion proteins, prodrugs, drugs in trials, previously approved drugs for AD, drugs developed for indications other than AD, and any fragments, derivatives, variants or complements of any of the above. Such agents can be used separately or in combination. Agents can be obtained from natural sources, such as, e.g., marine microorganisms, lgae, plants, and fungi. Alternatively, agents can be from combinatorial libraries of agents, including peptides or small molecules, or from existing repertories of chemical compounds synthesized in industry, e.g., by the chemical, pharmaceutical, environmental, agricultural, marine, cosmeceutical, drug, and biotechnological industries. Agents can include, e.g., pharmaceuticals, therapeutics, environmental, agricultural, or industrial agents, pollutants, cosmeceuticals, drugs, organic compounds, lipids, glucocorticoids, antibiotics, peptides, proteins, sugars, carbohydrates, and chimeric molecules.

Combinatorial libraries can be produced for many types of agents that can be 5 synthesized in a step-by-step fashion. Such agents include polypeptides, proteins, nucleic acids, beta-turn mimetics, polysaccharides, phospholipids, hormones, prostaglandins, steroids, aromatic compounds, heterocyclic compounds, benzodiazepines, oligomeric N- substituted glycines and oligocarbamates. Large combinatorial libraries of compounds can be constructed by the encoded synthetic libraries (ESL) method described in Affymax, WO

10 95/12608, Affymax WO 93/06121, Columbia University, WO 94/08051, Pharmacopeia, WO 95/35503 and Scripps, WO 95/30642 (each of which is incorporated herein by reference in its entirety for all purposes). Peptide libraries can also be generated by phage display methods. See, e.g., Devlin, WO 91/18980. Compounds to be screened can also be obtained from governmental or private sources, including, e.g., the National Cancer Institute's (NCI)

15 Natural Product Repository, Bethesda, MD, the NCI Open Synthetic Compound Collection, Bethesda, MD, NCI's Developmental Therapeutics Program, or the like.

The compounds also include several categories of molecules known to regulate gene expression, such as zinc finger proteins, ribozymes, siRNAs and antisense RNAs. Zinc finger proteins can be engineered or selected to bind to any desired target site within a gene of the

20 invention. An exemplary motif characterizing one class of these proteins (C2H2 class) is - Cys-(X)2-4-Cys-(X)12-His-(X)3-5-His (where X is any amino acid). A single finger domain is about 30 amino acids in length, and several structural studies have demonstrated that it contains an alpha helix containing the two invariant histidine residues and two invariant cysteine residues in a beta turn co-ordinated through zinc. In some methods, the target site is

25 within a promoter or enhancer. In other methods, the target site is within the structural gene. In some methods, the zinc finger protein is linked to a transcriptional repressor, such as the KRAB repression domain from the human KOX-I protein (Thiesen et al., New Biologist 2, 363-374 (1990); Margolin et al., Proc. Natl. Acad. Sci. USA 91, 4509-4513 (1994); Pengue et al., Nucl. Acids Res. 22:2908-2914 (1994); Witzgall et al., Proc. Natl. Acad. Sci. USA 91,

30 4514-4518 (1994)). In some methods, the zinc finger protein is linked to a transcriptional activator, such as VIPl 6. Methods for selecting target sites suitable for targeting by zinc finger proteins, and methods for design zinc finger proteins to bind to selected target sites are described in WO 00/00388. Methods for selecting zinc finger proteins to bind to a target using phage display are described by EP.95908614.1. The target site used for design of a zinc finger protein is typically of the order of 9-19 nucleotides.

Agents identified via these assays can be utilized to prevent, treat, diagnose and prognosticate AD-related disease. For example, when AD-related disease results from an overall lower level or activity of RNAs or polypeptides associated with resistance to AD- related disease, agents that enhance or stimulate the expression or activity of such RNAs or polypeptides may be used to treat or prevent AD-related disease. When AD-related disease results from the an overall higher level or activity of RNAs or polypeptides associated with susceptibility to AD-related disease, agents that inhibit or diminish the expression or activity of such RNAs or polypeptides may be used to treat or prevent AD-related disease. Optionally, an agent that modulates expression of a gene can be combined with an agent that modulates activity of a protein encoded by the gene. Optionally, agents that modulate expression of different genes which in combination result in AD-related disease can be combined. 1. Screening Assays For Agents That Modulate the Expression of Coding Nucleic Acids In one embodiment, agents that modulate (enhance, inhibit, or otherwise change) the level of expression of an AD polypeptide can be identified by comparing the level of expression of such coding nucleic acid in the presence of a test agent and in a control. A modulation of expression may occur at the DNA level (e.g., a transcription factor, etc.) or at the RNA level (antisense RNA, splicing, RNA-binding protein, etc.) A control can be in the absence of the test agent or a previously established level of expression. A solution or sample (e.g., cell or tissue culture) containing nucleic acids encoding an AD polypeptide can be contacted with a test agent. A solution can comprise, for example, cells or cell lysates containing the AD gene as well as other elements necessary for transcription/translation. Cells not suspended in solution as well as animal models may also be used. In addition, complexes of nucleic acid and protein agents may be detected by methods well known in the art. For example, such methods may utilize chromatography, microarrays, fluorescent labeling, and other methods further described in the section entitled Immobilization Assays herein.

If the level of expression of the AD coding nucleic acid is greater by an amount that is statistically significant from the level of expression in the control, then the test agent is an agonist of AD gene expression or activity. If the level of expression in the presence of the test agent is less by an amount that is statistically significant from the level of expression in the control, then the test agent is an antagonist of the expression of associated gene. The level of expression coding nucleic acids can be evaluated, for example, by determining the level of mRNA or polypeptides that are expressed, and/or any other method herein or known in the art, including but not limited to northern analysis, western blotting and antibodies. Using a similar method, agents that modulate the expression of associated gene variants associated with resistance or with susceptibility to AD-related disease can be identified. Preferably an agent is an agonist to the expression of associated genomic region variants associated with resistance to AD-related disease or an antagonist to the expression of associated genomic region variants associated with susceptibility to AD-related disease. More preferably, an agent is both an agonist to the expression of associated genomic region variants associated with resistance to AD-related disease and an antagonist of associated genomic region variants associated with susceptibility to AD-related disease.

2. Screening Assays For Agents That Modulate the Expression of Coding Nucleic

Acids By Interacting With Regulatory Regions

In another embodiment, agents that modulate the expression of coding AD nucleic acids by interacting with an AD regulatory region (e.g., enhancers, introns, 5' and 3' untranslated regions (e.g., promoters) and uORF's) are provided. For example, agents that modulate transcription or translation of nucleic acids herein (e.g., transcription factors) can be identified by contacting a solution containing non-coding nucleic acids associated with AD- related disease operably linked to a reporter gene with a test agent. After contact with the test agent, the level of expression of the reporter gene (e.g., the level of mRNA or polypeptide expressed) is assessed and compared with the level of expression in a control (e.g., the level of expression in the absence of a test agent or a level of expression that has previously been established). If the level of expression in the test sample is greater than the level of expression in the control sample by a statistically significant amount, then the test agent is an agonist of expression. If the level of expression in the test sample is less than the level of expression in a control sample by a statistically significant amount, then the test agent is an antagonist of the expression.

In some embodiments, an agent is an antagonist to the expression of associated genomic region variants associated with susceptibility to AD-related disease. In other embodiments, an agent is an agonist to the expression of associated genomic region variants associated with resistance to AD-related disease. In further embodiments, an agent is both an antagonist to the expression of AD variants associated with susceptibility to AD-related disease and an agonist to the expression of AD variants associated with resistance to AD- related disease. In particular embodiments, the agent increases the resistance and/or decreases the susceptibility of an organism (e.g., human) to AD-related disease by interacting with one or more AD regulatory nucleic acids.

3. Screening Assays For Agents That Enhance/Inhibit Polypeptide Activity

In another embodiment, agents that modulate (enhance, inhibit, or otherwise alter) the activity of polypeptides associated with AD-related disease (e.g., enhance the presence of certain splicing variants, or modulate one or more functions of the polypeptide (e.g., binding activity)) are identified by contacting a test agent with a cell, cell lysate or a solution containing nucleic acids and/or polypeptides associated with AD-related disease and comparing the activity of the polypeptides with their activity in a control (in absence of the test agent or a previously established level activity). If the activity of polypeptides associated with AD-related disease is enhanced by an amount that is statistically significant from the level of activity of the same polypeptides in a control, then the agent is an agonist of the activity of such polypeptides. If the activity of polypeptides associated with AD-related disease is inhibited by an amount that is statistically significant from the level of activity of the same polypeptides in a control, then the agent is an antagonist of the activity of such polypeptides. The activity of AD polypeptides may be modulated, e.g., by enhancing or inhibiting the expression of such polypeptides (i.e., increasing or decreasing the production of the polypeptides); by enhancing or inhibiting the activity of one or more such polypeptides (e.g., by altering the enzyme kinetics, binding affinity, etc. of the polypeptides); or by changing the cellular localization of one or more of such polypeptides. In a preferred embodiment, an agent is an agonist of the activity of polypeptides associated with resistance to AD-related disease. In another preferred embodiment, an agent is an antagonist of the activity of polypeptides associated with susceptibility to AD-related disease. Preferably, an agent is both an agonist of the activity of polypeptides associated with resistance to AD-related disease and an antagonist of the activity of polypeptides associated with susceptibility to AD-related disease.

4. Protein Agents That Bind AD Polypeptides

In another embodiment, assays can be used to identify protein agents that interact or bind one or more of the polypeptides herein, e.g., an AD polypeptide. Any method suitable for detecting protein-protein interactions may be employed for identifying protein agents that interact with or bind to AD polypeptides. Among the traditional methods that may be employed are co-immunoprecipitation, crosslinking, and co-purification through gradients or chromatographic columns.

In one embodiment, a yeast two-hybrid system, such as that described by Fields and Song (Fields, S. and Song, O., (1989) Nature 340:245-246), can be used to identify polypeptides that interact with one or more AD variants A yeast two-hybrid system employs two vectors The first vector has a DNA binding domain, the second, a transcription activation domain Each domain is fused to a sequence encoding a different polypeptide If the polypeptides interact with one another, transcriptional activation can be achieved, and 5 transcription of specific markers can be used to identify the presence of interaction and transcriptional activation In one example, a first vector contains a nucleic acid encoding a DNA binding domain and an AD polypeptide, and a second vector contains a nucleic acid encoding a transcription activation domain and test polypeptide which may potentially interact with the AD polypeptide (e g , a binding agent) Incubation of yeast containing the

10 first vector and the second vector under appropriate conditions (e g , mating conditions such as those used in the Matchmaker system from Clontech (Palo Alto, CA)) allows for the identification of colonies that express the markers of interest These colonies can be examined to identify the polypeptide(s) that interact with the AD polypeptide tested The binding molecules may be use as agents to alter the activity or expression of an AD

15 polypeptide as described above

In another embodiment, a protein microchip may be used to identify polypeptides that bind to AD polypeptides or any other polypeptide herein A protein microchip or microarray is provided having one or more protein complexes and/or antibodies selectively lmmunoreactive with a polypeptide of interest Protein microarrays are becoming

20 increasingly important in both proteomics research and protein-based detection and diagnosis of diseases The protein microarrays m accordance with this embodiment are be useful m a variety of applications including, e g , large-scale or high-throughput screening for compounds capable of binding to the protein complexes or modulating the interactions between the interacting protein members in the protein complexes

25 Protein microarrays can be prepared in a number of methods known in the art An example of a suitable method is that disclosed in MacBeath and Schreiber, (2002) Science, 289 1760-1763 Essentially, glass microscope slides are treated with an aldehyde-containing silane reagent (SuperAldehyde Substrates purchased from TeleChem International, Cupertino, Calif) Nanohter volumes of protein samples in a phosphate-buffered salme with

30 40% glycerol are then spotted onto the treated slides using a high-precision contact-printing robot After incubation, the slides are immersed in a bovine serum albumin (BSA)- contaming buffer to quench the unreacted aldehydes and to form a BSA layer that functions to prevent non-specific protein binding in subsequent applications of the microchip Alternatively, as disclosed in MacBeath and Schreiber, proteins or protein complexes of the present invention can be attached to a BSA-NHS slide by covalent linkages. BSA-NHS slides are fabricated by first attaching a molecular layer of BSA to the surface of glass slides and then activating the BSA with N,N'-disuccinimidyl carbonate. As a result, the amino groups of the lysine, aspartate, and glutamate residues on the BSA are activated and can form covalent urea or amide linkages with protein samples spotted on the slides. See MacBeath and Schreiber, Science, 289:1760-1763 (2000).

Another example of a useful method for preparing a protein microchip is disclosed in PCT Publication Nos. WO 00/4389A2 and WO 00/04382. First, a substrate or chip base is covered with one or more layers of thin organic film to eliminate any surface defects, insulate proteins from the base materials, and to ensure uniform protein array. Next, a plurality of protein-capturing agents (e.g., antibodies, peptides, etc.) are arrayed and attached to the base that is covered with the thin film. Proteins or protein complexes can then be bound to the capturing agents forming a protein microarray. The protein microchips are kept in flow chambers with an aqueous solution. The protein microarrays herein can also be made by the method disclosed in PCT

Publication No. WO 99/36576, which is incorporated herein by reference. For example, a three-dimensional hydrophilic polymer matrix, i.e., a gel, is first dispensed on a solid substrate such as a glass slide. The polymer matrix gel is capable of expanding or contracting and contains a coupling reagent that reacts with amine groups. Thus, proteins and protein complexes can be contacted with the matrix gel in an expanded aqueous and porous state to allow reactions between the amine groups on the protein or protein complexes with the coupling reagents thus immobilizing the proteins and protein complexes on the substrate. Thereafter, the gel is contracted to embed the attached proteins and protein complexes in the matrix gel. The protein microchips of the present invention can also be prepared with other methods known in the art, e.g., those disclosed in U.S. Pat. Nos. 6,087,102, 6,139,831, 6,087,103; PCT Publication Nos. WO 99/60156, WO 99/39210, WO 00/54046, WO 00/53625, WO 99/51773, WO 99/35289, WO 97/42507, WO 01/01142, WO 00/63694, WO 00/61806, WO 99/61148, WO 99/40434, all of which are incorporated herein by reference. 5. Agents That Interfere with AD Interaction with Binding Agents

The polypeptides herein may interact in vivo with one or more cellular or extracellular binding agents (e.g., polypeptides, nucleic acids, etc.) to form a complex. Agents that disrupt such an interaction may be used to regulate the activity or function of the AD polypeptides herein. Such agent may include, but are not limited to molecules such as antibodies, peptides, and the like. Assays that assess the impact of a test agent on the activity of an AD polypeptide in relation to a cellular or extracellular binding agent are provided. These assays involve the preparation of a reaction mixture containing an AD polypeptide and a cellular or extracellular binding agent and a time sufficient to allow the two products to interact and bind thus forming a complex.

To test an agent for inhibitory activity, reaction mixtures are prepared in the presence and absence of the test agent. The test agent can be initially included in the reaction mixture or added at a time subsequent to the addition of the AD polypeptide and/or its cellular or extracellular binding agent. Control reaction mixtures can be incubated without the test agent or with a placebo agent. Formation of complexes between AD polypeptides and cellular or extracellular binding agents is measured in both in the control and test reaction mixtures. A difference in the formation of a complex in the control reaction and the test reaction mixture indicates that the compound affects the interaction of the AD polypeptide and the cellular or extracellular binding agent. For example, the agent may enhance or inhibit binding between the AD-related disease polypeptide and the binding agent. Additionally, complex formation in a reaction mixture containing a test agent and AD polypeptide may be compared to complex formation in a reaction mixture containing the test agent and a second AD polypeptide that is encoded by a different nucleic acid sequence than the first AD polypeptide. In certain embodiments, the first and second AD polypeptides are encoded by different alleles of the same gene. This comparison can be important in those cases in which it is desirable to identify agents that disrupt interaction of a particular AD polypeptide.

The screening assays for test agents that interfere with AD polypeptide interaction with binding agents may be conducted in a heterogeneous or homogeneous format. Heterogeneous assays involve anchoring one of the binding partners onto a solid phase and detecting complexes anchored on the solid phase at the end of the reaction. In homogeneous assays, the entire reaction is carried out in a liquid phase. In either approach, the order of addition of reactants can be varied to obtain different information about the agents being tested. In either example, test agents that affect the interaction between the AD polypeptides and the cellular or extracellular binding agents can be tested, for example, by competition by adding the test agent to the reaction mixture prior to, post, or simultaneously within the AD polypeptide and cellular or extracellular binding agents and assessing the difference in complex formation. Alternatively, test agents that disrupt or otherwise affect formed complexes, (e.g., compounds (e.g., with higher binding constants) that displace one of the components from the complex) can be tested by adding the test agent to the reaction mixture after the complexes have been formed.

The ability or effectiveness of a test agent to bind to an AD polypeptide, a cellular or extracellular binding agent, or a complex thereof can be assessed, for example, by coupling a test agent with a radioisotope or enzymatic label such that binding of the test agent to the AD polypeptide, binding agent, or complex thereof can be determined by detecting the label (e.g., 1251, 35S, 14C, or 3H) either directly or indirectly (e.g., by direct counting of radio emission or by scintillation counting). Alternatively, test agents can be enzymatically labeled with, for example, horseradish peroxidase, alkaline phosphatase or luciferase and the enzymatic label can be detected by determination of conversion of an appropriate substrate to a product.

In another embodiment, the ability of a test agent to interact with an AD polypeptide, binding agent, or complex thereof can be assessed without the labeling of any of the interactants. For example, a microphysiometer can be used to detect the interaction of a test agent with an AD polypeptide or a binding agent without the labeling of either the test agent, the AD polypeptide or the binding agent. See McConnell, H.M. et al. (1992) Science 257:1906-1912. As used herein, a "microphysiometer" (e.g., Cytosensor (Molecular Devices, Sunnyvale, CA)) is an analytical instrument that measures the rate at which a cell acidifies its environment using a light-addressable potentiometric sensor (LAPS). Changes in this acidification rate can be used as an indicator of the interaction between the binding agent and the AD polypeptide.

6. Screening For Small Molecules

Agents that enhance or inhibit the expression, function, and/or activity of AD nucleic acids or polypeptides can be obtained using any of the numerous approaches in combinatorial library methods known in the art, including: biological libraries; natural products libraries; spatially addressable parallel solid phase or solution phase libraries; synthetic library methods requiring deconvolution; the 'one-bead one-compound' library method; and synthetic library methods using affinity chromatography selection. The biological library approach is largely limited to polypeptide libraries, while the other four approaches are applicable to polypeptide, non-peptide oligomer or small molecule libraries of compounds. See Lam, K.S. (1997) Anticancer Drug Des. 12:145.

Non-peptide agents or small molecules are generally preferred because they are more readily absorbed after oral administration and have fewer potential antigenic determinants. Small molecules are also more likely to cross the blood brain barrier than larger protein-based pharmaceuticals. Methods for screening small molecule libraries for candidate protein-binding molecules are well known in the art and may be employed to identify molecules that modulate (e.g., through direct or indirect interaction) one or more of the AD polypeptides herein. Briefly, AD polypeptides may be immobilized on a substrate and a solution including the small molecules is contacted with the AD polypeptide under conditions that are permissive for binding. The substrate is then washed with a solution that substantially reflects physiological conditions to remove unbound or weakly bound small molecules. A second wash may then elute those compounds that are bound strongly to the immobilized polypeptide. Alternatively, the small molecules can be immobilized and a solution of AD polypeptides can be contacted with the column, filter or other substrate on which the small molecules are immobilized. The ability to detect binding of an AD polypeptide to a small molecule may be facilitated by labeling (e.g., radio-labeling or chemiluminescence) the polypeptide or small molecule.

In another embodiment, electronic molecular modeling applies an algorithm to screen small molecule databases for ligands and molecules that interact or bind with AD polypeptides or those in pathways therewith. See Meng et al., (1992) J. Comp. Chem.

15:505. In one example the DOCK3.5 is used to screen for small molecules that interact with AD polypeptides, preferably the binding pocket of an AD polypeptide. A "negative image" of the binding pocket on a protein surface is created. The image is created by the computational equivalent of placing atom-sized spheres into the binding pocket. A representative set of spheres are identified by DOCK3.5 that fit extremely well into the binding pocket. The generated spheres constitute an irregular grid that is matched to the atomic centers of potential ligands. The list of atom centers, or more conveniently the matrix of interatomic distances linking these atom centers forms a useful description of the binding site. The matrix of interatomic distances for the putative ligand is also made. The best mutual overlap of the two matrices is sought. This alignment specifies the orientation of the ligand relative to the negative image of the protein and thus docks the ligand into the protein's binding pocket.

Non-peptide agents or small molecule libraries can be prepared by a synthetic approach, but recent advances in biosynthetic methods using enzymes may enable one to prepare chemical libraries that are otherwise difficult to synthesize chemically. Small molecule libraries can also be obtained from various commercial entities, for example, SPECS and BioSPEC B.V. (Rijswijk, the Netherlands), Chembridge Corporation (San Diego, California), Comgenex USA Inc., (Princeton, N.J.), Maybridge Chemical Ltd. (Cornwall, U.K.), and Asinex (Moscow, Russia). These small molecule libraries can be screening in a high throughput manner to identify one or more agents For example, a high throughput screening assay for small molecules that was disclosed in Stockwell, B R et al , Chem & Bio , (1999) 6 71-83, is a miniaturized cell-based assay for monitoring biosynthetic processes such as DNA synthesis and post-translational processes 7 Immobilization Assays

In any embodiment herein, it may be desirable to immobilize either the AD polypeptides, the test agent or other components of the assay (e g , binding agents) on a substrate to facilitate the separation of bound polypeptides from unbound polypeptides, as well as to accommodate automation of the assay A substrate can be any vessel suitable for containing the reactants Examples of substrates include microtrter plates, test tubes, and micro-centπfuge tubes In one example, agents that bind a polypeptide of interest can be detected by anchoring either the polypeptide of interest (e g , any polypeptide herein) or the test agent (e g , antibody) to a substrate (e g , microtiter plates) and then detecting complexes of the polypeptide of interest and test agent anchored to the substrate at the end of the reaction Where the polypeptide of interest is anchored and the test agent is not anchored, the test agent can be labeled, either directly or indirectly In other embodiments, the polypeptide or other components of the assay maybe labeled, either directly or indirectly

In a preferred embodiment, microtiter plates are used as the solid phase, and the anchored component can be immobilized by non-covalent or covalent attachments Non- covalent attachments can be achieved by simply coating the solid surface with a solution of the protein and drying In another preferred embodiment, an immobilized antibody (preferably a monoclonal antibody) specific for the polypeptide to be immobilized can be used to anchor the polypeptide to the solid surface The surface can be prepared in advance and stored In another embodiment, a fusion protein (e g , a glutathione-S-transferase fusion protein) can be provided which adds a domain that allows the polypeptides, binding agents or test agents to be bound to a matrix or other solid support A non-immobilized component is then added to the coated surface containing the anchored component After the reaction is complete, unreacted components are removed (e g , by washing) and complexes anchored on the solid surface are detected Where the non-immobilized component is pre-labeled, the detection of label immobilized on the surface indicates that the complexes were formed Where the non-immobilized component is not pre-labeled, an indirect label can be used to detect complexes anchored on the surface, such as by using a labeled antibody specific for the non-immobilized component The antibody can then be labeled or indirectly labeled, e g , with an anti-Ig antibody.

Alternatively, this reaction can be conducted in a liquid phase, the reaction products separated from unreacted components, and complexes detected using, for example, an immobilized antibody specific for a polypeptide of interest or test agent to anchor the complexes formed in solution and a labeled antibody specific for the other component of the possible complex to detect anchored complexes.

In another embodiment, an assay performed in liquid phase has the pre-formed complexes of the AD polypeptides and the cellular or extracellular binding agents prepared such that either the polypeptide or the binding agents are labeled, but the signal generated from the label is eliminated or diminished due to complex formation. The addition of a test agent that competes with and displaces one of the species from the pre-formed complex results in the generation of a signal above background.

In one particular embodiment, the AD polypeptide is prepared using recombinant DNA techniques described herein and is fused to a glutathione-S-transferase (GST) gene using a fusion factor such as pGEX-5X- 1 , such that its binding activity is maintained in the resulting fusion product. The cellular or extracellular binding agent is purified and used to raise a monoclonal antibody, using methods routinely practiced in the art. This antibody can be labeled with the radioactive isotope 1251, for example by methods known in the art. In a substrate binding assay, the GST-AD polypeptide fusion product is anchored, for example, to glutathione-agarose beads. The cellular or extracellular binding agent is then added in the presence or absence of the test agent in a manner that allows interaction and binding to occur. At the end of the reaction period, unbound material is washed away, and the labeled monoclonal antibody can be added to the system and allowed to bind to the complexed components. The interaction between the AD polypeptide and the cellular or extracellular binding agents is detected by measuring the amount of radioactivity that remains associated with the beads. A successful inhibition of the interaction by the test agent will result in a decrease in measured radioactivity.

Alternatively, the GST bound AD polypeptide fusion product and the interactive cellular or extracellular binding agent can be mixed together in liquid in the absence of the solid glutathione-agarose beads. The test agent is added either during or after the binding agent is allowed to interact with the GST-fusion polypeptide. This mixture is then added to the glutathione-agarose beads and unbound material is washed away. The extent of inhibition of the binding agent interaction can be detected by adding the labeled antibody and comparing the radioactivity associated with the beads to that of a control reaction (e.g., lacking test agent)

The same techniques can also be employed using polypeptide fragments, deπvatives, or variants that correspond to the binding domains of either the AD polypeptides (e g , BH3) or the cellular or extracellular binding agents, or both Binding sites can be identified and isolated using any one of a number of methods known in the art, including for example site directed mutagenesis

Alternatively, an AD polypeptide can be anchored to a solid substrate using methods disclosed herein and allowed to interact with and bind its labeled binding agent, which has been previously treated with a proteolytic enzyme (e g , trypsin) After washing, a short- labeled peptide comprising the binding domain (e g , BH3) remains associated with the solid material, which can be isolated and identified by amino acid sequencing Also, once the gene coding for the cellular or extracellular binding agent is obtained, short gene segment can be engineered to express bindmg fragments, which can then be tested for binding activity, purified and/or synthesized 8 Agents That Enhance/Inhibit Genes in The AD Pathways

AD may further be prevented or treated by administering to a patient an agent that enhances or inhibits the expression or activity of genes in the associated gene pathways Genes in the associated gene pathways are those that act upstream or downstream of the associated genomic regions m an AD-related pathway, and whose gene products may interact with, bind to, compete with, induce, enhance, or inhibit, directly or indirectly, the activity, expression, or function of genes in the associated genomic regions, or any gene whose gene products are downstream of associated genomic regions, wherein the associated genomic region induces, enhances or inhibits the expression of activity of such gene products, directly or indirectly Genes in the pathways of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUROG3, C10ORF35, LOC729099, ND3, ND4, and homologs thereof are particularly contemplated by the present invention Also contemplated by the present invention are genes in the pathway of, PSEN2, APP, HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNF113B, EXOC2, LOC642335, HDAC4, ZNF366, LOC389300, LOC644154, LOC645932, and homologs thereof

9 Potential Agents And Binding Sites

Agents that modulate the expression or activity of AD polypeptides include nucleic acids, transcription factors, antisense nucleic acids, polypeptides, fusion proteins, PNAs, mimetics (e g , soluble peptides or Ig-tailed fusion peptides), antibodies (e g , monoclonal, polyclonal, humanized, anti-idiotypic, chimeric or smgle-cham antibodies, Fab, F(ab')2, Fab expression library fragments, and epitope-bmdmg fragments thereof), binding molecules, prodrugs, drugs in trials, previously approved drugs, drags developed for indications other than AD-related disease, small and large organic or inorganic molecules, and any fragments, derivatives, variants or complements of any of the above Such agents may be used separately or in combination

Any of the agents herein can also serve as "lead agents" in the design and development of new pharmaceuticals For example, sequential modification of small molecules (e g , amino acid residue replacement with peptides, functional group replacement with peptide or non-peptide compounds) is a standard approach in the pharmaceutical industry for the development of new pharmaceuticals Such development generally proceeds from a lead agent, which is shown to have at least some of the activity of the desired pharmaceutical In particular, when one or more agents having at least some activity of interest are identified, structural comparison of the molecules suggest portions of the lead agents that should be conserved and portions that may be varied in the design of new candidate compounds This embodiment also encompasses means of identifying lead agents that may be sequentially modified to produce new candidate agents for use in the treatment of AD-related disease These new agents may be tested for therapeutic efficacy (e g , in the cell- based or animal models described herein) This procedure may be iterated until compounds having the desired therapeutic activity and/or efficacy are identified 10 Cell Based Assays and Animal Models

The invention provides transgenic animals having a genome comprising a transgene comprising an exogenous nucleic acid encoding a protein encoded by a gene selected from the group consisting of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, APOC4, BCAM, LOC728050, NEUR0G3, C10ORF35, LOC729099, ND3, ND4, PSEN2, APP,

HDAC4, OLFMl, RAB12, KIAA0802, CLYBL, ZIC5, LOC728155, LOC727827, FARPl, RNFl 13B, EXOC2, LOC642335, ZNF366, LOC389300, LOC644154, and LOC645932 The exogenous nucleic acid can be genomic, cDNA or mimgene The exogenous nucleic is usually from another species, particularly human, but can also be from the same species, in which case it occupies a different genomic location than the corresponding endogenous gene

The nucleic acid preferably includes the susceptibility allele at a polymorphic site shown in Table E, and in certain embodiments a susceptibility allele shown in Table 19-1 The coding sequence of the gene is in operable linkage with regulatory element(s) required for its expression Such regulatory elements can include a promoter, enhancer, one or more mtrons, πbosome binding site, signal sequence, polyadenylation sequence, 5' or 3' UTR and 5' or 3' flanking sequences The regulatory sequence can be from the gene being expressed or can be heterologous If heterologous, the regulatory sequences are usually obtained from a gene known to be expressed in the intended tissue in which the gene of the invention is to be expressed (e g , the CNS) For example, regulatory sequences from a prion gene, PDGF or Thy-1 are suitable The transgenic animals are disposed to develop at least one sign or symptom of an AD-related disease, particularly LOAD

The invention also provides transgenic animals in which a nonhuman homolog of one of the human genes of the invention is disrupted or enhanced so as to so as to reduce, eliminate or increase its expression relative to a nontransgemc animal of the same species Disruption can be achieved either by genetic modification of the nonhuman homolog or by functional disruption by introducing an inhibitor of expression of the gene (as discussed above) into the nonhuman animal Enhancement of expression can be achieved either by genetically modifying the regulatory element(s) associated with an endogenous gene (e g , introducing a stronger promoter), or by using a zinc finger protem linked to an appropriate activating domain, as discussed above

Some transgenic animals have a plurality of transgenes respectively comprising a plurality of genes of the invention Some transgenic animals have a plurality of disrupted nonhuman homologs of genes of the invention Some transgenic animals combine both the presence of transgenes expressing one or more genes of the invention and one or more disruptions of nonhuman homologs of other genes of the invention

Transgenic animals of the invention are preferably rodents, such as mice or rats, or insects, such as Drosophila Other transgenic animals such as pπmates, ovines, porcmes, caprmes and bovmes can also be used The transgene in such animals is integrated into the genome of the animal The transgene can be integrated in single or multiple copies Multiple copies are generally preferred for higher expression levels In a typical transgenic animal all germline and somatic cells include the transgene in the genome with the possible exception of a few cells that have lost the transgene as a result of spontaneous mutation or rearrangement For some animals, such as mice and rabbits, fertilization is performed in vivo and fertilized ova are surgically removed In other animals, particularly bovmes, it is preferable to remove ova from live or slaughterhouse animals and fertilize the ova in vitro See DeBoer et al , WO 91/08216 Methods for culturing fertilized oocytes to the pre-implantation stage are described by Gordon et al , Methods Enzymol 101, 414 (1984), Hogan et al , Manipulation of the Mouse Embryo A Laboratory Manual, C S H L N Y (1986) (mouse embryo); Hammer et al., Nature 315, 680 (1985) (rabbit and porcine embryos); Gandolfi et al. J. Reprod. Fert. 81, 23-28 (1987); Rexroad et al., J. Anim. Sci. 66, 947-953 (1988) (ovine embryos) and Eyestone et al. J. Reprod. Fert. 85, 715-720 (1989); Camous et al., J. Reprod. Fert. 72, 779-785 (1984); and Heyman et al. Theriogenology 27, 5968 (1987) (bovine embryos) (incorporated by reference in their entirety for all purposes). Sometimes pre- implantation embryos are stored frozen for a period pending implantation. Pre-implantation embryos are transferred to the oviduct of a pseudopregnant female resulting in the birth of a transgenic or chimeric animal depending upon the stage of development when the transgene is integrated. Chimeric mammals can be bred to form true germline transgenic animals. Alternatively, transgenes can be introduced into embryonic stem cells (ES). These cells are obtained from preimplantation embryos cultured in vitro. Bradley et al., Nature 309, 255-258 (1984) (incorporated by reference in its entirety for all purposes). Transgenes can be introduced into such cells by electroporation or microinjection. ES cells are suitable for introducing transgenes at specific chromosomal locations via homologous recombination. Transformed ES cells are combined with blastocysts from a non-human animal. The ES cells colonize the embryo and in some embryos form or contribute to the germline of the resulting chimeric animal. See Jaenisch, Science, 240, 1468-1474 (1988) (incorporated by reference in its entirety for all purposes).

Alternatively, transgenic animals can be produced by methods involving nuclear transfer. Donor nuclei are obtained from cells cultured in vitro into which a human alpha synuclein transgene is introduced using conventional methods such as Ca-phosphate transfection, microinjection or lipofection. The cells are subsequently been selected or screened for the presence of a transgene or a specific integration of a transgene (see WO 98/37183 and WO 98/39416, each incorporated by reference in their entirety for all purposes). Donor nuclei are introduced into oocytes by means of fusion, induced electrically or chemically (see any one of WO 97/07669, WO 98/30683 and WO 98/39416), or by microinjection (see WO 99/37143, incorporated by reference in its entirety for all purposes). Transplanted oocytes are subsequently cultured to develop into embryos which are subsequently implanted in the oviducts of pseudopregnant female animals, resulting in birth of transgenic offspring (see any one of WO 97/07669, WO 98/30683 and WO 98/39416).

For production of transgenic animals containing two or more transgenes, the transgenes can be introduced simultaneously using the same procedure as for a single transgene. Alternatively, the transgenes can be initially introduced into separate animals and then combined into the same genome by breeding the animals. Alternatively, a first transgenic animal is produced containing one of the transgenes. A second transgene is then introduced into fertilized ova or embryonic stem cells from that animal. Optionally, transgenes whose length would otherwise exceed about 50 kb, are constructed as overlapping fragments. Such overlapping fragments are introduced into a fertilized oocyte or embryonic stem cell simultaneously and undergo homologous recombination in vivo. See Kay et al., WO 92/03917 (incorporated by reference in its entirety for all purposes).

Nonhuman homologs of human genes of the invention can be disrupted by gene targeting. Gene targeting is a method of using homologous recombination to modify a mammalian genome, can be used to introduce changes into cultured cells. By targeting a gene of interest in embryonic stem (ES) cells, these changes can be stably introduced into the germline of laboratory animals. The gene targeting procedure is accomplished by introducing into tissue culture cells a DNA targeting construct that has a segment that can undergo homologous combination with a target locus and which also comprises an intended sequence modification (e.g., insertion, deletion, point mutation). The treated cells are then screened for accurate targeting to identify and isolate those which have been properly targeted. A common scheme to disrupt gene function by gene targeting in ES cells is to construct a targeting construct which is designed to undergo a homologous recombination with its chromosomal counterpart in the ES cell genome. The targeting constructs are typically arranged so that they insert additional sequences, such as a positive selection marker, into coding elements of the target gene, thereby functionally disrupting it. Similar procedures can also be performed on other cell types in combination with nuclear transfer. Nuclear transfer is particularly useful for creating knockouts in species other than mice for which ES cells may not be available Polejaeva et al., Nature 407, 86-90 (2000)). Breeding of nonhuman animals which are heterozygous for a null allele may be performed to produce nonhuman animals homozygous for said null allele, so-called "knockout" animals (Donehower et al. (1992) Nature 256: 215; Science 256: 1392, incorporated herein by reference).

Any of the compositions herein can be tested for their ability to prevent, ameliorate, delay, or treat symptoms associated with AD-related disease, especially dementia and memory loss. Cell-based systems can be useful for identifying agents that ameliorate symptoms associated with AD-related disease. Cell-based systems include cells that express one or more of the AD polypeptides herein and exhibit cellular phenotypes associated with resistance or susceptibility to AD-related disease. Cell-based systems include recombinant transgenic cell lines derived from animals containing one or more cells expressing one or more of the nucleic acids herein. Preferably, such cells provide a continuous cell line. CeIl- based systems also include non-recombinant cell lines preferably from primary tissues of patients having AD-related disease or resistance to AD-related disease

A cell-based system having a phenotype of AD-related disease can be exposed to an agent suspected of ameliorating phenotypic states associated with susceptibility to AD-related disease at a sufficient concentration and for a time sufficient to elicit such an amelioration response in the exposed cells After exposure, the cells can be examined to determine whether the phenotypic states have been altered such that the phenotype has been eliminated and the cells resemble normal phenotypes or phenotypes of resistance to AD-related disease Ammal models can be used to determine toxicity, efficacy and/or mechanism of action of the agents identified herein Animal models for AD-related disease include both non-recombmant and recombmant transgenic animals Non-recombmant animal models for AD-related disease include, for example, dog and murine models Murine models can be created, for example, by administering to an ammal an effective amount of alcohol or a drug to elicit a response or symptom associated with AD-related disease Such animal models can then be exposed to an agent suspected of ameliorating AD-related disease

Additionally, recombmant ammal models exhibiting phenotypic states of AD- related disease or resistance thereto can be engineered, for example, by introducing nucleic acids associated with susceptibility or resistance, respectively In specific embodiments, recombmant animal models can be engineered to exhibit early or late age-of-onset of AD- related disease, for example, by introducing nucleic acids associated with early or late age-of- onset, respectively In one embodiment, an engineered sequence includes at least part of the target nucleic acid sequence and disrupts the endogenous target sequence upon integration of the engineered target gene sequence into the animal's genome Techniques for making a transgenic animal are known in the art For example, target gene sequences may be introduced into, and overexpressed in, the genome of the animal of interest, or, if endogenous AD-related gene sequences are present, they may either be overexpressed or, alternatively, be disrupted to underexpress or inactivate AD-related gene expression, such as described for the disruption of apoE in mice (Plump et al (1992) Cell 71 343-353) Other techniques include, for example, pronuclear microinjection disclosed m U S Pat No 4,873,191, retrovirus mediated gene transfer into germ-lmes disclosed in Van der Putten et al , (1985) Proc Natl Acad Sci USA, 82 6148-6152, gene targeting in embryonic stem cells disclosed in Thomson et al , (1989) Cell 56 313-321 , electroporation of embryos disclosed in Lo, (1983) MoI Cell Biol (3) 1803-1814, and sperm-mediated gene transfer disclosed in Lavitrano et al, (1989) Cell 57 717-723, etc For a review of such techniques, see Gordon (1989) Transgenic Animals, Intl. Rev. Cytol. 115:171-229. Nucleic acids can also be introduced into some, but not all cells of an animal to create a mosaic animal. Selective introduction into and activation of a particular cell type is discussed, for example, in Lasko et al. (1992) Proc. Natl. Acad. Sci. USA 89:6232-6236. An engineered sequence includes preferably at least part of the target nucleic acid sequence. This disrupts the endogenous target sequence upon integration of the engineered target gene sequence into the animal's genome.

In a preferred embodiment, the nucleic acids herein are used to over-express polypeptides associated with resistance to AD-related disease. In another preferred embodiment, the nucleic acids herein are used to underexpress polypeptides associated with susceptibility to AD-related disease. To overexpress a polypeptide, for example, a nucleic acid encoding the polypeptide of interest can be ligated to a regulatory sequence that can drive the expression of the polypeptide in the animal cell type of interest. Such regulatory regions are well known. In another example, a non-genic nucleic acid (e.g., an intron or a regulatory sequence) may be introduced alone to drive the production of a polypeptide of interest. To underexpress an endogenous polypeptide, a nucleic acid encoding a transcription factor that down-regulates the polypeptide or a nucleic acid that produces a variant or inactive polypeptide may be introduced into the genome of an animal such that the endogenous expression will be inactivated. In addition to, or in the alternative, a non-genic nucleic acid herein (e.g., an intron nucleic acid) may be introduced separately to override native regulatory region.

Any of the animal models disclosed herein can be used to identify agents capable of ameliorating, treating, delaying, or preventing symptoms associated with susceptibility to AD-related disease. For example, animal models can be exposed to a compound suspected of exhibiting an ability to ameliorate one or more symptoms associated with AD-related disease at a sufficient concentration and for a time sufficient to elicit an ameliorating response in the exposed animal. The response of the exposed animal can be monitored by assessing change in symptoms. Any treatments that diminish one or more symptoms associated with AD- related disease or susceptibility thereto may be considered a candidate for human therapy. Dosages of test agents can be determined by deriving dose-response curves, which are well- known and commonly used in the art.

VI Pharmaceutical Compositions

Any of the agents and compositions identified herein may be produced in quantities sufficient for pharmaceutical administration and/or testing.

Pharmaceutical compositions can be formulated in accordance with the routine procedures adapted for administration to human beings. Often, pharmaceutical compositions are formulated with an acceptable carrier or excipient. See Remington's Pharmaceutical Sciences, Gennaro, A., (ed., Mack Publishing Co. 1990).

Suitable pharmaceutically acceptable carriers include but are not limited to water, salt solutions (e.g., NaCl), saline, buffered saline, alcohols, glycerol, ethanol, gum arabic, vegetable oils, benzyl alcohols, polyethylene glycols, gelatin, carbohydrates such as lactose, amylose or starch, dextrose, magnesium stearate, talc, silicic acid, viscous paraffin, perfume oil, fatty acid esters, hydroxymethylcellulose, polyvinyl pyrolidone, etc., as well as combinations thereof.

Pharmaceutically acceptable salts include those formed with free amino groups such as those derived from hydrochloric, phosphoric, acetic, oxalic, tartaric acids, etc., and those formed with free carboxyl groups such as those derived from sodium, potassium, ammonium, calcium, ferric hydroxides, isopropylamine, triethylamine, 2 ethylamino ethanol, histidine, procaine, etc.

The pharmaceutical compositions can include, if desired, auxiliary agents, e.g., lubricants, preservatives, stabilizers, wetting agents, emulsifiers, salts for influencing osmotic pressure, buffers, coloring, flavoring and/or aromatic substances and the like which do not deleteriously react with the active agents.

The pharmaceutical compositions, if desired, can also contain minor amounts of wetting or emulsifying agents, or pH buffering agents. The composition can be a liquid solution, suspension, emulsion, tablet, pill, capsule, sustained release formulation, or powder. The composition can be formulated as a suppository, with traditional binders and carriers such as triglycerides.

The pharmaceutical compositions and their physiologically acceptable salts and solvates can be formulated for administration by inhalation or insufflation (either through the mouth or the nose, or oral, buccal, parenteral, or rectal administration). For administration by inhalation, the compositions are conveniently delivered in the form of an aerosol spray presentation from pressurized packs or a nebulizer, with the use of a suitable propellant, e.g., dichlorodifluoromethane, trichlorofluoromethane, dichlorotetrafluoroethane, carbon dioxide, or other suitable gas. In the case of a pressurized aerosol, the dosage unit can be determined by providing a valve to deliver a metered amount. Capsules and cartridges of, e.g., gelatin for use in an inhaler or insufflator can be formulated containing a powder mix of the compound and a suitable powder base such as lactose or starch.

For oral administration, the pharmaceutical compositions can take the form of tablets or capsules prepared by conventional means with pharmaceutically acceptable excipients such as binding agents, fillers, disintegrants, or wetting agents, sweeteners, including, pregelatinised maize starch, polyvinylpyrrolidone, hydroxypropyl methylcellulose, fillers, lactose, microcrystalline cellulose, calcium hydrogen phosphate, lubricants, magnesium stearate, talc, silica, potato starch or sodium starch glycolate, sodium lauryl sulfate, mannitol, lactose, starch, magnesium stearate, polyvinyl pyrollidone, sodium saccharine, cellulose and magnesium carbonate. The tablets can be coated by methods well known in the art. Preparations for oral administration can be suitably formulated to give controlled release of the active compound.

Liquid preparations for oral administration can take the form of solutions, syrups, or suspensions, or they can be presented as a dry product for constitution with water or other suitable vehicle before use. Such liquid preparations can be prepared by conventional means with pharmaceutically acceptable additives such as suspending agents, e.g., sorbitol syrup, cellulose derivatives, or hydrogenated edible fats; emulsifying agents, e.g., lecithin or acacia; non-aqueous vehicles, e.g., almond oil, oily esters, ethyl alcohol, or fractionated vegetable oils; and preservatives, e.g., methyl or propyl-p-hydroxybenzoates or sorbic acid. The preparations can also contain buffer salts, flavoring, coloring, and/or sweetening agents as appropriate.

In particular, the liquid preparations can be administered in a beverage. Such beverage can be alcoholic, non-alcoholic beverage or a health beverage. Such beverage may comprise one or more of the agents or compositions herein as well as, optionally, any one or more of the following: alcohol fructose, vitamins, electrolyte substitutes, caffeine, amino acids, minerals, artificial and natural sweeteners, milk or dry-milk powder and other additives and preserving agents.

Examples of vitamins that may be included are components of the vitamin B complex, such as vitamin Bl, B2, B6, B12, biotin, niacin, pantothenic acid, folic acid, adenine, choline, adenosine phosphate, orotic acid, pangamic acid, carnitine, 4-aminobenzoic acid, myo-inositol, liponic acid and/or amygdaline. In the body, vitamin Bl, also known as thiamin, is converted into thiamin-pyrophosphate, a coenzyme in a number of reactions in which C-C bonds are cleaved. It can also be added as thiamin hydrochloride. Vitamin B2, also known as riboflavin, is reabsorbed in the small intestines, converted into FMN (flavin mononucleotide) and, in the liver, into FAD (flavin-adenine-dinucleotide), both of which are coenzymes in redox reactions Vitamin B6, also known as pyndoxal, pyrodoxin and pyπdoxamme, is a component of pyndoxal-5 -phosphate, which is a co factor m glycogen degradation and in ammo acid metabolism, e g as a coenzyme of decarboxylases Preferably, this substance is admixed into the beverage in the form of pyridoxin hydrochloride Vitamin B 12, also known as cyanocobalamine, has a complex structure and is a component of cobalamme-coenzymes, with methyl-cobalamine and cobalamide, e g , being involved in rearrangements with hydrogen migration Biotin, also known as vitamin B7, is covalently bound to carboxylases Niacin, also known as B3, is a generic name for nicotinic acid and nicotinamide Niacin is a component of NAD and its phosphate, NADP, and is one of the most important hydrogen transmitters in the cell having a protective and anabolic effect on the body Pantothenic acid, also known vitamin B3 or B5, has a precursor function for coenzyme A which assumes a central position m metabolism Folic acid, or vitamin B9, is a component of the coenzyme tetrahydrofolate. Vitamin C may further be provided

Preferably, the beverage composition comprises components of the vitamin B complex in the following parts by weight, based on a total of 15,000-20,000 parts by weight of the dry substance vitamin Bl, 0 1-10 parts by weight, preferably 1 part by weight, vitamin B2, 0 1-10 parts by weight, preferably 1 5 parts by weight, vitamin B6, 0 1-10 parts by weight, preferably 1 5 parts by weight, biotin, 0 01-1 parts by weight, preferably 0 1 parts by weight, niacin, 0 1-100 parts by weight, preferably 10-30 parts by weight, pantothenic acid, 0 1-100 parts by weight, preferably 1 - 10 parts by weight, vitamin B 12, 0 0001 -0 1 parts by weight, preferably 0 001-0 01 parts by weight, folic acid, 0 01-10 parts by weight, preferably 0 1 parts by weight, and/or vitamin C, 0 1-500 parts by weight, preferably 50 parts by weight

It is advantageous for the beverage to comprise ammo acids, in particular L- glutamine and/or L-argimne Amino acids play an important role in the various metabolic processes of the human body In particular, L-glutamine and L-arginme may be admixed in the beverage according to the following parts by weight, based on a total of 15,000-20,000 parts by weight of dry substance L-argimne, 20-2,000 parts by weight, preferably 200 parts by weight, and/or L-glutamine, 10-1,000 parts by weight, preferably 100 parts by weight Caffeine is optionally added at 0 1-1 OO parts by weight, preferably 25 parts by weight, based on a total of 15,000-20,000 parts by weight

Examples of minerals that may be used include magnesium, potassium, zinc and calcium In particular, potassium and magnesium play an important role in metabolism and are involved m many ATP-catalyzed enzyme reactions A mineral may be added separately, in combination, and/or in combination with other food additives, e g as magnesium

IOO glycerophosphate, potassium citrate (acid regulator), zinc gluconate (fruit acid) and calcium pantothenate. Minerals are preferably added at the following parts by weights, based on a total of 15,000-20,000 parts by weight of the dry substance: magnesium, 10-1,000 parts by weight, preferably 100 parts by weight; potassium 10-1,000 parts by weight, preferably 100 parts by weight; zinc, 0.1-100 parts by weight, preferably 10 parts by weight; calcium 10- 1,000 parts by weight, preferably 100 parts by weight.

A tastier beverage may further include sugars and/or artificial sweeteners. Both artificial and natural sweeteners may be added to sweeten the compositions herein. Besides fructose, any other sugar may be admixed, such as glucose, galactose, lactose, etc. Artificial sweeteners include, for example, aspartame, saccharine and cyclamate as well as any other commercially available artificial sweeteners.

Furthermore, the compositions herein may comprise of further additives, in particular flavoring agents, preserving agents, coloring agents, antioxidants, electrolytes, enzymes, plant extracts, glycerolphosphates, acid regulators and/or acidifiers, in particular fruit acids.

A beverage may be carbonated or non-carbonated, and may be combined or based on liquids such as fruit juices, milk, tea, coffee, water etc. Moreover, alcohol may be admixed to the beverage herein.

The compositions can be formulated for intravenous administration. Compositions used for intravenous administration are typically solutions in sterile isotonic aqueous buffer. Where necessary, the compositions may also include a solubilizing agent and a local anesthetic to ease pain at the site of the injection. Generally, the ingredients are supplied either separately or mixed together in unit dosage form, for example, as a dry lyophilized powder or water free concentrate in a hermetically sealed container such as an ampule indicating the quantity of active agent. Where the composition is to be administered by infusion, it can be dispensed with an infusion bottle containing sterile pharmaceutical grade water, saline or dextrose/water. Where the compositions are administered by injection, an ampule of sterile water for injection or saline can be provided so that the ingredients may be mixed prior to administration. The compositions can be formulated for parenteral administration by injection, e.g., by bolus injection or continuous infusion. Formulations for injection can be presented in unit dosage form, e.g., in ampoules or in multi-dose containers, with an added preservative. The compositions can take such forms as suspensions, solutions, or emulsions in oily or aqueous vehicles, and can contain formulatory agents such as suspending, stabilizing, and/or dispersing agents. Alternatively, the active ingredient can be in powder form for constitution with a suitable vehicle, e.g., sterile pyrogen-free water, before use.

For topical application, nonsprayable forms, viscous to semi-solid or solid forms comprising a carrier compatible with topical application and having a dynamic viscosity preferably greater than water, can be employed. Suitable formulations include but are not limited to solutions, suspensions, emulsions, creams, ointments, powders, enemas, lotions, sols, liniments, salves, aerosols, etc., which are, if desired, sterilized or mixed with auxiliary agents, e.g., preservatives, stabilizers, wetting agents, buffers or salts for influencing osmotic pressure, etc. The agent may be incorporated into a cosmetic formulation. For topical application, also suitable are sprayable aerosol preparations wherein the active ingredient, preferably in combination with a solid or liquid inert carver material, is packaged in a squeeze bottle or in admixture with a pressurized volatile, normally gaseous propellant, e.g., pressurized air.

The compounds can also be formulated in rectal compositions such as suppositories or retention enemas, e.g., containing conventional suppository bases such as cocoa butter or other glycerides.

In addition to the formulations described previously, the compounds can also be formulated as a depot preparation. Such long acting formulations can be administered by implantation (for example, subcutaneously or intramuscularly) or by intramuscular injection. Thus, for example, the compounds can be formulated with suitable polymeric or hydrophobic materials (for example as an emulsion in an acceptable oil) or ion exchange resins, or as sparingly soluble derivatives, for example, as a sparingly soluble salt.

The compositions can, if desired, be presented in a pack or dispenser device that can contain one or more unit dosage forms containing the active ingredient. The pack can for example comprise metal or plastic foil, such as a blister pack. The pack or dispenser device can be accompanied by instructions for administration. Pharmaceutical packs or kits comprising one or more containers filled with one or more of the ingredients of the pharmaceutical compositions disclosed herein are also provided. Optionally, associated with such containers can be a notice in the form prescribed by a governmental agency regulating the manufacture, use or sale of pharmaceuticals or biological products, which notice reflects approval by the agency of manufacture, use of sale for human administration. The packs or kits can be labeled with information regarding mode of administration, sequence of drug administration (e.g., separately, sequentially or concurrently) or the like. The packs or kits may also include means for reminding the patient to take the therapy. The packs or kits can comprise of a single unit dosage of the combination therapy or a plurality of unit dosages. In particular, the compositions can be separated, mixed together or present in a single vial or tablet. Compositions assembled in a blister pack or other dispensing means are preferred. Unit dosages provided are preferably dependent on the pharmacodynamics of each agent and administered in FDA approved dosages in standard time courses. VII Methods For Treatment

The agents and pharmaceutical compositions herein can be used as prophylactic or therapeutic treatment of AD-related disease. AD-related disease may result from excessive levels of certain gene products (e.g., AD polypeptides) or deficient levels of other gene products (e.g., polypeptides associated with resistance to AD-related disease).

1. Indications for Treatment

Some indications that may be used to diagnose certain AD-related diseases, e.g., LOAD, include memory loss for simple things like friends' names, commonly used phone numbers, or what month it is and how to get to a familiar place; misplacement of things more often than usual; loss of train of thought when speaking; repeating things often; feeling more suspicious, cautious, or anxious; loss of interest in things that used to be enjoyable; and feeling of stress when making decisions. A more definitive indication of AD-related disease may require a biopsy.

2. Methods for Administration The agents and pharmaceutical compositions herein can be administered separately or in combination, in an amount effective to treat an indication of interest. For example, a patient diagnosed with or afflicted by AD-related disease may be administered a therapeutically effective amount of an inhibitor of polypeptides associated with susceptibility to AD-related disease to reduce the level of activity and/or expression of such polypeptides. In the alternative, a patient diagnosed with or afflicted by an AD-related disease may be administered a therapeutically effective amount of an agonist of polypeptides associated with resistance to AD-related disease to reduce the level of activity and/or expression of such polypeptides. More preferably, a patient diagnosed with or afflicted by an AD-related disease is administered a combination treatment of both inhibitors of polypeptides associated with susceptibility to AD-related disease and agonists of polypeptides associated with resistance to AD-related disease. Such combination treatment may require lower dosages due to the synergetic effect of both compounds.

Examples of agents that may be administrated in combination with any of the compositions herein to treat or prevent AD-related disease include: cholinesterase inhibitors (e.g., galantamine, rivastigmine, donepezil, and tacrine); N-methyl D-aspartate (NMDA) antagonist (e.g., memantine); common heart drugs (e.g., ACE inhibitors), cholesterol- lowering drags (e.g., Lipitor™ and Zocor™), galantamine hydrobromide, gingko biloba, and cholinesterase inhibitors (ChEIs) (e.g., donepezil), and other agents described above.

The agents and pharmaceutical compositions may be administered or coadministered orally, parenterally, intraperitoneally, intravenously, intraarterially, transdermally, sublingually, intramuscularly, rectally, transbuccally, intranasally, liposomally, via inhalation, vaginally, intraoccularly, via local delivery (for example by catheter or stent), subcutaneously, intraadiposally, intraarticularly, or intrathecally. The compounds and/or compositions may also be administered or co-administered in slow release dosage forms. Other suitable methods include gene therapy using rechargeable or biodegradable devices, particle acceleration devices ("gene guns") and slow release polymeric devices. The pharmaceutical compositions herein can also be administered as part of a combinatorial therapy with other agents.

The combination of therapeutic agents and compositions may be administered by a variety of routes, and may be administered or co-administered in any conventional dosage form. Co-administration in the context of this invention is defined to mean the administration of more than one therapeutic in the course of a coordinated treatment to achieve an improved clinical outcome. Such co-administration may also be coextensive, that is, occurring during overlapping periods of time. For example, an associated genomic region antisense may be administered to a patient before, concomitantly, or after the administration of an inhibitor of AD polypeptides.

In a preferred embodiment, a pharmaceutical compound is administered orally, and more preferably is self-administered. For example, a beverage comprising one or more agents or pharmaceutical compositions may be administered to prevent, ameliorate or treat AD-related disease. The dosage of active ingredients may be based on the composition, its interaction with other compounds, or the stage of progression of AD-related disease in a patient, for example. A patient is preferably monitored following administration of an agent of the invention for changes in a sign or symptom of the AD-related disease against which treatment or prophylaxis is being effected.

3. Gene Replacement Therapy In another embodiment, nucleic acids can be introduced into recipient cells using techniques such as gene replacement therapy.

Preferably, one or more nucleic acids associated with resistance to AD-related diesease may be inserted into appropriate cells within a patient, using vectors such as adenovirus, adeno-associated virus and retrovirus vectors. Nucleic acids can also be introduced into cells via particles, such as liposomes. Other techniques for direct administration involve stereotactic delivery of such sequences to the site of the cells in which the sequences are to be expressed.

Methods for introducing nucleic acids into mammalian cells are well known in the art. Generally, the nucleic acid is directly administered in vivo into a target cell or a transgenic mouse that expresses SP-10 promoter operably linked to a reporter gene. This can be accomplished by any methods known in the art, e.g., by constructing it as part of an appropriate nucleic acid expression vector and administering it so that it becomes intracellular, e.g., by infection using a defective or attenuated retroviral or other viral vector (U.S. Pat. No. 4,980,286), by direct injection of naked DNA, by use of microparticle bombardment (e.g., a gene gun; Biolistic, Dupont), by coating with lipids or cell-surface receptors or transfecting agents, by encapsulation in liposomes, microparticles, or microcapsules, by administering it in linkage to a peptide which is known to enter the nucleus, or by administering it in linkage to a ligand subject to receptor-mediated endocytosis (Wu and Wu, (1987) J. Biol. Chem. 262:4429-4432), which can be used to target cell types specifically expressing the receptors. In another embodiment, a nucleic acid-ligand complex can be formed in which the ligand comprises a fusogenic viral peptide to disrupt endosomes, allowing the nucleic acid to avoid lysosomal degradation. In yet another embodiment, the nucleic acid can be targeted in vivo for cell specific uptake and expression, by targeting a specific receptor (see, e.g., PCT Publications WO 92/06180 dated Apr. 16, 1992; WO 92/22635 dated Dec. 23, 1992; WO92/20316 dated Nov. 26, 1992; WO93/14188 dated JuI. 22, 1993; WO 93/20221 dated Oct. 14, 1993).

Additional methods that may be utilized to increase or decrease the overall level of expression of an AD nucleic acid include using targeted homologous recombination methods to modify the expression characteristics of an endogenous sequence in a cell or microorganism by inserting a heterologous DNA regulatory element such that the inserted regulatory element is operatively linked with the endogenous sequence in question. Targeted homologous recombination can thus be used to activate transcription of an endogenous nucleic acid that is transcriptionally silent, (e.g., not normally expressed or expressed at very low levels), to silence the transcription of an endogenous nucleic acid that is transcriptionally active, or to enhance or decrease the expression of an endogenous sequence that is normally expressed.

Further, the overall level of expression of polypeptides associated with resistance to AD may be increased by the introduction of cells that express such polypeptides associated with resistance to AD, preferably autologous cells, into a patient at positions and in numbers which are sufficient to prevent or ameliorate symptoms or conditions associated with AD- related disease. Such cells may be either recombinant or non-recombinant. In a preferred embodiment, such cells are healthy brain cells. When the cells to be administered are non-autologous cells, they can be administered using well-known techniques that prevent a host immune response against the introduced cells from developing. For example, the cells may be introduced in an encapsulated form that, while allowing for an exchange of components with the immediate extracellular environment, does not allow the introduced cells to be recognized by the host immune system.

The amounts of therapeutic agents or compositions to be administered can vary, according to determinations made by one of skill, but preferably are in amounts effective to create reduce or reverse AD-related disease symptoms. Treatment compositions and dosages can be specifically tailored to each situation based on an individual patient's pharmacogenomics (response to a drug), phenotype, genotype and the compositions used for treatment. Preferably, for co-administration, the total amounts are less than the total amounts for each pharmaceutical compound added together. For the slow-release dosage form, appropriate release times can vary, but preferably should last from about 1 hour to about 6 months, most preferably from about 1 week to about 4 weeks. Formulations for slow release dosage can vary as determinable by one of skill, according to the particular situation and as generally taught herein.

The LD50 (the lethal dose to 50% of the population) and the ED50 (the effective dose in 50% of the population) of a pharmaceutical composition can be determined using cell cultures or animal models following standard pharmaceutical procedures. The dose ratio of lethal and effective doses is the therapeutic index and is expressed as the ratio LD50/ED50 Compounds that exhibit large therapeutic indices are preferred. Compounds that exhibit toxic side effects can also be used, but care should be taken to design a delivery system that targets such compounds to the site of affected tissue to minimize potential damage to uninfected cells. When using cell culture to estimate the therapeutically effective dose, the dosage of such compounds lies preferably within a range of circulating concentrations that include the ED50 with little or no toxicity. A dose can also be formulated in animal models to achieve a circulating plasma concentration range that includes the IC50 (the concentration of the test 5 compound that achieves a half-maximal inhibition of symptoms) as determined in cell culture. Such information can be used to more accurately determine useful doses in humans. Levels in plasma can be measured, for example, by high performance liquid chromatography.

The combination of therapeutic agents may be used in the form of kits. The arrangement and construction of such kits is conventionally known. Such kits may include 10 containers for containing the inventive combination of therapeutic agents and/or compositions, and/or other apparatus for administering the inventive combination of therapeutic agents and/or compositions.

Throughout the disclosure various patents, patent applications and publications are referenced. Unless otherwise indicated, each is incorporated by reference in its entirety for 15 all purposes. All patents, patent applications, and publications mentioned herein are cited for the purpose of describing and disclosing reagents, methodologies and concepts that may be used in connection with the present invention. Nothing herein is to be construed as an admission that these references are prior art in relation to the inventions described herein. The invention will be further described by the following non-limiting examples. 20

EXAMPLES

Example 1 - Overview of Association Study

LOAD is a devastating neurodegenerative disease, characterized by the formation of pathogenic plaques in the brain, which affects as many as 10% of people 65 or older. The

25 disease is complex, and is likely to involve the interaction of a number of genes.

Apolipoprotein E (APOE) has been reported to increase the risk of LOAD and lower the age of onset (Corder, E. H. et al. Science 261, 921-3 (1993); Fairer, L. A. et al. Jama 278, 1349- 56 (1997); Saunders, A. M. et al. Neurology 43, 1467-72 (1993)). In addition, linkage peaks have been reported on chromosomes 9, 10, and 12 ( Kehoe, P. et al. Hum MoI Genet 8, 237-

30 45 (1999); Pericak- Vance, M. A. et al. Jama 278, 1237-41 (1997); Pericak- Vance, M. A. et al. Exp Gerontol 35, 1343-52 (2000); Ertekin-Taner, N. et al. Science 290, 2303-4 (2000); Bertram, L. et al. Science 290, 2302-3 (2000); Blacker, D. et al. Hum MoI Genet 12, 23-32 (2003); Myers, A. et al. Am J Med Genet 114, 235-44 (2002)). This Example identifies genetic loci associated with LOAD. A whole-genome association study was performed in a total of 800 unrelated case and control samples, using a dense set of SNP markers (approximately 1.5 million) that cover the entire genome. To individually screen such a large number of SNPs in all 800 samples would be prohibitive. Therefore, a two-stage approach was used. First a screen for associations using pooled sample sets was performed. Allele frequency differences between the case and control sample pools was estimated, and the estimated allele frequency differences were used to select a subset of SNPs for further evaluation. This subset, which may have contained false positives in addition to true positives, was genotyped in the individual case and control samples, and the exact allele frequency differences between the populations was calculated. SNPs showing significant association with LOAD in the original sample set were analyzed in a second sample set, to verify their association.

For Phase I our aim was to demonstrate the effectiveness of this two-stage design for identifying SNPs associated with LOAD by using a subset of the 1.5 million SNP set. Overall, we analyzed approximately 250,000 of the 1.5 million SNPs in the pooled samples. We then selected approximately 20,000 of these SNPs and genotyped them in the individual case and control samples. In addition, we genotyped almost 5,000 SNPs from candidate regions thought to contain loci associated with LOAD (including APOE). Example 2 - Scanning the Entire Human Genome The entire human genome was scanned to identify common variants (and others) using microarray technology platforms such as described in US 6,969,589; U.S. Ser. No. 10/284,444, entitled "Chromosome 21 SNPs, SNP Groups and SNP Patterns," filed on October 31, 2002, assigned to the same assignee as the present application; and US 6,897,025, all of which are incorporated herein by reference. The microarrays are manufactured using a process adapted from semiconductor manufacturing to achieve cost effectiveness and high quality.

Example 3 - Haplotvpe Blocks

Variants identified were grouped into haplotype blocks using methods disclosed in U.S. Ser. Nos. 10/106,097, entitled "Methods for Genomic Analysis", filed March 26, 2002, incorporated herein by reference. Representative variants and haplotype blocks from an entire human chromosome (chromosome 21) are disclosed in, for example, Patil, N. et al, "Blocks of Limited Haplotype Diversity Revealed by High-Resolution Scanning of Human Chromosome 21" Science 294, 1719-1723 (2001) and the associated supplemental materials, and Hinds, et al. "Whole-Genome Patterns of Common DNA Variation in Three Human Populations" Science 307, 1072 (2005), both of which are incorporated herein by reference. In particular, the preparation of a linkage disequilibrium map of the entire human genome is reviewed by Hinds et al., Science 307, 1072-79 (Feb. 2005), accompanying on-line material available at genome.perlegen.com. For brevity, the SNPs in linkage disequilibrium with each other are sometimes referred to in this application (and the scientific literature) as occupying a linkage disequilibrium bin. As such, other SNPs within the same linkage disequilibrium bin as an exemplified SNP cosegregate together as a result of linkage disequilibrium and are thus expected to have the same association with AD-related disease. Identification of additional mutations beyond those provided herein that are in linkage disequilibrium with the SNPs provided herein would be the inevitable result of repeating the analysis of Hinds at higher resolution on more SNPs.

Example 4 - Samples

Four hundred unrelated case samples were obtained from individuals with late-onset Alzheimer's Disease having an apparent familial basis (i.e., each individual was from a family in which at least two members had the disease but only one member per family was included in the sample). Two hundred eighty-eight of these were obtained from the NIMH Center for Genetics Initiative. The NIMH collection contains multiplex families ascertained with two or more living related individuals with AD. Individuals were evaluated clinically, with longitudinal follow-up, and diagnoses were confirmed at autopsy. We used the following criteria to select a subset of individuals with LOAD. The individuals had to be: (1) classified as "definite" (confirmed on autopsy) or "probable" (consensus or clinical diagnosis) AD; (2) age of onset for all affected members of the family > 65; and (3) Caucasian. Only one individual (usually the proband) was chosen from each family. We obtained an additional 112 unrelated case samples from the Department of

Veterans Affairs, Seattle, WA. These samples were selected using the same criteria as for the NIMH samples. A diagnosis of AD had been confirmed for all of these samples at autopsy, and thus all were classified as "definite" AD. Table 4-1 shows the characteristics of the 400 case samples used in this study.

Table 4-1 : Characteristics of the 400 case samples used

Total samples Definite AD Probable AD Average age Age range

Female 257 141 116 74.5 65-93

Male 143 63 80 74.4 65-90 Total 400 204 196

Four hundred cognitively-normal control samples were obtained from Group Health Cooperative, Seattle, as part of the UW/Group Health Alzheimer's Disease Patient Registry (UOl AG 06781 - 18). They were selected from a larger set of community-based samples collected from Caucasian enrollees of Group Health (a Seattle Area HMO) aged 65 or older. Patients with no prior diagnosis of dementia were assessed by the Cognitive Abilities Screening Instrument (CASI), which is a 100-point scale based on the MMSE and Hasegawa DRS and 3MS tests. Patients were assessed every two years and those who continue to score more than 86 (equivalent to about 26 on the MMSE test) and did not show any symptoms of dementia were classified as controls.

Example 5 - Assay Design for Pooled Genotvping

We designed assays for a total of 267,852 SNPs to be examined in the sample pools. These SNPs are distributed relatively evenly across the genome, with an average gap size of 11 kb. Thus, these SNPs are not focused on specific regions of the genome. We have extensive experience with this set of high-quality SNPs, all of which have minor allele frequencies of at least 10% in Caucasian populations and can be genotyped with our technology.

We used oligonucleotide arrays designed such that each SNP was interrogated by forty distinct 25 bp probes. These forty features consisted of four sets often features, corresponding to the forward and reverse strands of the two SNP alleles (reference and alternate). Each set often features consisted of two sets of five features, with offsets of -2, -1, 0, +1, and +2 bases between the center of the 25 bp probe and the SNP position. For each offset, we tiled one perfect-match feature and one mismatch feature (complement of the perfect match) at the central position of the probe. Thus, for each allele there were a total of ten perfect-match probes and ten mismatch probes. The oligonucleotide features necessary to query the 267,852 SNPs studied here were arrayed on three distinct array designs. Example 6 - Generation of Pools

Samples were analyzed for quality as follows: (1) concentration and volume were measured to make sure that they matched the expected values, and were adequate for the study; (2) gel electrophoresis was performed on a subset of samples to examine DNA integrity; and (3) PCR assays were performed to establish the ability of a subset of the DNA samples to be amplified.

After passing QC the samples were diluted to a concentration of 600μg/ml, and re- quantified by PicoGreen assay. We divided the samples into a total of eight pools, four containing case samples and four containing control samples. Thus, each pool contained 100 samples, randomly selected from either the cases or controls, with each sample present in just one pool. Equimolar amounts (600 ng) of each sample were transferred into one of the eight pools robotically. Each pool was then re-quantified by PicoGreen assay and diluted to a standard concentration for use as a PCR template. Example 7 - Genotyping of pooled samples

The pools were independently amplified using multiplexed PCR with a single primer pair for each SNP. The amplified products were pooled, labeled and hybridized to the three different chip designs that together query the set of 267,852 SNPs described above. The hybridized chips were washed and stained with Cy-chrome. The hybridization of labeled sample was detected by measuring Cy-chrome fluorescence.

After removing SNP measurements that failed quality control (see below), the estimated allele frequency difference between case and control pools, termed delta p-hat, was automatically derived for each SNP from intensity ratios for hybridization to the allele- specific 25-mer features. The fluorescence intensities of the reference and alternate perfect- match features on the arrays correlate with the concentration of the corresponding SNP allele in the DNA sample. Our estimates of allele frequency, p-hat, were computed from ratios of trimmed means of intensities of the perfect-match features, after subtracting a measure of background computed from trimmed means of intensities of mismatch features. The case pool ρ-hats and control pool p-hats were separately averaged, and the delta p-hat was calculated. Finally, the standard error of the estimate, based on the within pool variance of the measurements-, t-statistic p-value, and empirical p-values (which were obtained as rank of T_TEST_P_VALUE on each chip design divided by the total number of passing SNP measurements for each chip design) for the delta p-hat were calculated for each of the SNPs that passed the QC filters. Example 8 - Quality control

The following quality control filters were applied to the data to assess the reliability of the fluorescence intensities of the features for each SNP in an array scan. Applying these filters, which are based on findings from numerous previous association studies, increases the quality of the passing SNPs, thereby reducing false-positive associations. SNP measurements were removed from consideration if they had any of the following: (1) conformance of < 0.9; (2) saturated probes; and (3) signal-to-background ratio of <1.5.

The conformance of alleles was defined as the fraction of feature pairs for which the perfect-match feature is brighter than the corresponding mismatch feature. A conformance of < 0.9 can indicate the absence of target DNA. Both saturated probes and low signal-to- background ratios can lead to unreliable p-hat measurements.

Due to technical problems, PCR plates containing 970 SNPs did not amplify and were therefore not analyzed in the pools. While there were no pooled data for these SNPs, they were included in the individual genotyping (see below). Table 8-1 shows that the pass- rate for the remaining 266882 SNPs is 93.80%, with the eight case and control pools producing consistently high pass-rates.

Table 8-1 : Pooled genotyping SNP pass-rates

Pool Passed SNPs Assayed SNPs Pass-rate case pool 1 253653 266882 95.04 case pool 2 246856 266882 92.50 case pool 3 247842 266882 92.87 case pool 4 249290 266882 93.41 average case pool 93.45 control pool 1 251434 266882 94.21 control pool 2 252618 266882 94.66 control pool 3 252309 266882 94.54 control pool 4 248703 266882 93.19 average control pool 94.15

Overall average 93.80

Example 9 - Selection of SNPs for further evaluation

SNPs were selected for further evaluation (individual genotyping) if they fell into one of the following categories:

(1) SNPs were selected if they met both of the following criteria in the pooled genotyping: (a) the corrected empirical p- values for the delta p-hat were < 0.08; and (b) measurements passed QC filters in at least two of the four case pools and at least two of the four control pools;

(2) SNPs were selected if they met both of the following criteria in the pooled genotyping: (a) the empirical p- values for the delta p-hat were < 0.0015; and

(b) measurements passed QC filters in at least two of the four case pools and at least two of the four control pools;

(3) SNPs were selected if they met all three of the following criteria in the pooled genotyping: (a) measurements passed QC filters in at least three of the four case pools and at least three of the four control pools; (b) the standard error (SE) of the delta p-hat measurements is < 0.04; and (c) delta p-hat > 0.07;

(4) SNPs not amplified in the pooled genotyping phase;

(5) Genomic control SNPs; or

(6) SNPs from candidate regions (see Table 9-2 for details of regions studied) The number of SNPs that fell into each category is given in Table 9- 1. Note that due to the inherent experimental noise of the pooled genotyping, the majority of the category 1 -3 SNPs will be false positives. True associations will be identified after genotyping this set of SNPs in individual case and control samples.

Table 9-1 : Selection of SNPs by category

Category Number of SNPs selected

1 20,125

2 26 (another 350 overlapped category 1)

3 0 (another 2930 overlapped category 1)

4 970

5 307 (another 4 overlapped categories 1-4)

6 4562 (another 317 overlapped category 1-5) total 25990

Table 9-2: Additional SNPs chosen from candidate regions

Chromosome Interval Start End SNPs picked for SNP size position position genotyping density

10 35 Mb 60,000,000 95,000,000 3656 1/9.6 kb 12 8 Mb 2,000,000 10,000,000 1173 1/6.8 kb 19 200 kb 50,000,000 50,200,000 50 1/4 kb total 4879

SNPs genotyped in individual samples

A total of 25,990 SNPs were individually genotyped in each of the 800 case and control samples. These included SNPs selected on the basis of the pooled genotyping results, SNPs from the candidate regions, and genomic control SNPs (see Tables 9-1 and 9-2). High-density oligonucleotide arrays

We created a new array design to genotype the selected SNPs, such that all 25,990 SNPs would be assayed using a single chip for each individual DNA sample. Example 10 - Individual genotyping of case and control samples

We used multiplexed PCR to amplify the SNPs from the 800 individuals. We pooled the PCR reactions from a single individual together, and created one labeled target from each PCR pool. We hybridized the 800 individuals (targets) to oligonucleotide arrays, thereby querying each SNP once in each sample. The hybridized chips were washed and stained, and the resulting fluorescence detected as for the pooled genotyping.

Individual genotypes for each SNP were determined by clustering the intensity measurements of all samples, in the two-dimensional space defined by background-adjusted trimmed mean intensities of the perfect-match features for the reference and alternate alleles. See Hinds, D. A. et al. Matching strategies for genetic association studies in structured populations. Am J Hum Genet 74, 317-25 (2004); Hinds, D. A. et al. Application of pooled genotyping to scan candidate regions for association with HDL cholesterol levels. Human Genomics 1, 421-34 (2004); and Hinds, D. A. et al. Whole genome patterns of common DNA variation in diverse human populations. Science (Submitted). We used a K-means algorithm to assign the measurements to clusters representing the three distinct diploid genotypes that are possible: homozygous-reference, heterozygous, and homozygous-alternate. The K-means and background optimization steps were iterated until cluster membership and background estimates converged. To determine the appropriate number of genotype clusters, we repeated the analysis for 1, 2, and 3 clusters, and selected the most likely solution, considering likelihoods of the data and the cluster parameters.

Using the STRUCTURE program with the genotyping results for the 311 genomic control SNPs, we then tested for population stratification and computed the genomic control variance inflation factor. For the SNPs that passed our quality control criteria, we then computed trend test and genomic control-adjusted trend test p-values. The trend test tests for the difference in genotypes (11, 12 and 22) between cases and controls. It provides a chi- square distributed statistic that does not rely on the Hardy- Weinberg equilibrium of the SNP (contrary to a chi-square test based on allele counts rather than genotypes). The test is additive in the sense that if Allele 2 is the predisposing allele, the individual with a 12 genotype is at half the increased risk of an individual with a 22 genotype. This is in contrast to models that are recessive (only 22 is predisposed) or dominant (the increased risk of an individual with a 12 genotype is equal to the risk of a 22 individual). The genomic control- adjusted trend test was corrected using the variance inflation factor computed using genomic control markers. Quality control The following quality control filters were applied to the data to assess the reliability of the fluorescence intensities of the features for each SNP in an array scan: a call-rate of 0.8, meaning that the SNP has an unambiguous genotype call in at least 80% of the samples; and a Hardy- Weinberg equilibrium p-value of > 0.0001. SNP call rates were computed after discarding genotypes that obtained < 0.2 score with our IG quality metric. The metric uses machine learning algorithm to approximate a probability of a genotype being discordant with outside platforms from 15 QC and SNP -property based inputs. Applying these filters, which are based on findings from numerous previous association studies, increases the quality of the passing SNPs. A total of 23,319 SNPs (90%) passed the individual genotyping quality filters, and were analyzed further.

Individual genotyping results: genetic stratification

The individuals used in this study are all self-reported Caucasians, which should limit the potential for population stratification issues. However, to test for possible stratification between the case and control sample sets, we used the clustering algorithm STRUCTURE (Pritchard, J. K. & Rosenberg, N. A. Use of unlinked genetic markers to detect population stratification in association studies. Am J Hum Genet 65, 220-8 (1999)) to analyze the genotypes of 311 genomic control SNPs in the individual case and control samples. The genomic control SNPs are distributed roughly evenly across the autosomes (Hinds, D. A. et al. Matching strategies for genetic association studies in structured populations. Am J Hum Genet 74, 317-25 (2004)). Figure 10-1 shows that when the samples were tested for two distinct populations, the cases and controls showed similar distributions of inferred population memberships. However, there is evidence of some difference in the distribution of the case and control individuals within the population. We used the genomic control trend test (Bacanu, S. A., Devlin, B. & Roeder, K. The power of genomic control. Am J Hum Genet 66, 1933-44 (2000)), which takes into account the average differences between case and control samples in the distribution, to correct for this. Individual genotyping results Assessing the false-positive and false-discovery rates

We calculated the number of SNPs with significant trend test p-values that we would expect to find purely by chance, assuming no enrichment of large allele frequency differences in the pooling phase of the study. From the 23,319 SNPs that passed individual genotyping quality filters, we expect to find 23,319 x p-value cutoff false-positives. So, for a trend test p-value of less than 1.00E-05, we expect to find 23,319 x 1.00E-05, or 0.23319 false-positives. As shown in Table 10-1, the number of observed SNPs below each of the different p-value cutoffs was well above the expected number This indicates that the pooled genotyping did indeed enrich our SNP set for SNPs with large allele frequency differences

Using the results from the 200 kb Chromosome 19 candidate region, for which 44 SNPs passed our quality filters, we estimated the false discovery rate We calculated the false-discovery rate as the ratio of the expected number of false positives to the number of observed SNPs with significant trend test p-values below a certain cutoff As shown in Table 10-2, the false-discovery rates estimated for different trend test p-value cutoffs are extremely low e g for SNPs with trend test p-values of less than 1 00E-08, the false discovery rate is estimated to be 6 29E-08

Assessing the power of our pooled genotypme method

The aim of the pooled genotyping is to enrich for significant SNPs, so that we reduce the number of SNPs requiring genotyping in each of the individual samples We expect a large number of the SNPs selected for individual genotyping on the basis of the pooled genotypmg results to be false positives Thus it is important to show that amongst these false positives we have captured a large fraction of the SNPs that we are looking for - those with true allele frequency differences between the case and control populations. That is, is the pooled genotyping powerful enough to identify SNPs with significant allele frequency differences between the case and control pools, even if the differences are relatively small. In this experiment we used criteria that resulted in the selection of approximately

8% of the SNPs that were genotyped in the pooled samples for follow-up genotyping in the individual samples. This was done because our previous studies had suggested that it would provide strong enough power to identify truly-associated SNPs while limiting the number of false-positives. The analysis of SNPs within the three candidate regions by both individual and pooled genotyping gave us the opportunity to actually quantify the power of the pooled genotyping in this study.

A total of 3591 SNPs were genotyped in both the pooled samples and the individual samples. The individual genotyping results showed that 1181 of these had allele frequency differences of > 0.03, with 9 having allele frequency differences of > 0.08 and 3 having allele frequency differences of > 0.1 (see Table 10-3). Table 10-3 shows the percentage of the SNPs with differential allele frequencies that would be selected for individual genotyping when different selection criteria are used. As expected, the ability to detect SNPs with differences in allele frequencies using pooled genotyping depends on the size of the difference, with the smallest differences being the most difficult to identify. In addition, selecting more of the SNPs leads to more of the significant SNPs being selected, e.g. selection of 20% of the SNPs analyzed by pooled genotyping would lead to 88.89% of the SNPs with allele frequency differences of at least 0.08 being identified, whereas selection of 2% would lead to only 44.44% being identified. However, as most of the 20% will be false- positives, we have to balance the potential to identify important SNPs with the expense of genotyping large numbers of false positives. The results show that by using the criteria which resulted in the selection of 8% of the SNPs genotyped in the pools we identified approximately 56% of the SNPs with an allele frequency difference of at least 0.08 (see highlighted column in Table 10-3). Thus, the pooled genotyping strategy used in this study provided enough power to identify a large fraction of the SNPs with the relatively small allele frequency differences likely to be important for the complex disease of LOAD.

Table 10-3: Power of pooled genotyping method to identify SNPs with different allele frequencies

Figure 10-2 shows the distribution of genomic control-corrected trend test p- values for the SNPs selected from throughout the genome by pooled genotyping (a), and the SNPs from the Chromosome 10 (b), 12 (c), and 19 (d) candidate regions. The results show that of the SNPs selected on the basis of the pooled genotyping results, there are many more than expected by chance with significant p-values. This indicates that genotyping the SNPs in the pooled samples before individual genotyping leads to an enrichment of SNPs with significant allele frequency differences, as intended. In contrast, the numbers of SNPs with significant p-values in the Chromosome 10 and 12 candidate regions are no larger than those expected by chance. For the Chromosome 19 candidate region there are a number of SNPs with highly significant p-values, as discussed in the next section.

Example 11 Candidate regions

Three chosen candidate regions were examined in detail by individual genotyping, using a high density of SNPs (see Table 9-2)

Chromosome 19 APOE region

A total of 44 SNPs from the 200 kb region surrounding the APOE gene on Chromosome 19 passed the individual genotyping quality control and were examined for association (23 of these were also genotyped in the pooled samples) This region has been studied in some detail by others, including Martm et al (Martin, E R et al SNPmg away at complex diseases analysis of single-nucleotide polymorphisms around APOE in Alzheimer disease Am J Hum Genet 67, 383-94 (2000), who examined a 1 5 Mb region surrounding APOE, and found association of SNPs located withm 40 kb of either side of APOE We genotyped a total of 44 SNPs in a similar interval around the APOE gene Fig 1 shows the results of the individual genotyping for all of these SNPs (see USSN 11/344,975, incorporated herein by reference) It also shows the pooled genotyping results for the 23 SNPs that were included in the pooled genotyping (the remaimng 21 were only individually genotyped because of their location m candidate regions) Eight of the SNPs showed significant allele frequency differences between the case and control samples (highlighted in green) One of the significant SNPs is in the intron of APOE, the gene well known to contribute to LOAD, another is in an intron of the polio virus receptor-related 2 gene (PVRL2), six more within or in close proximity with the translocase of outer mitochondrial membrane 40 homolog gene (TOMM40) Thus, we have found a number of highly- significant SNPs in the area surrounding the APOE gene

Although the majority of the SNPs analyzed in this region were specifically included in the individual genotyping because we wanted to provide high-density coverage of this important region, the results show that all seven of the significant SNPs that were genotyped in the pooled samples would have been selected on the basis of the pooled genotyping results alone Only one of the significant SNPs (rsl 6979513) was not included in the pooled genotyping screen The gene containing this SNP, TOMM40, was identified by other significant SNPs that were genotyped in the pools These results highlight the success of our two-stage association study strategy

Chromosomes 10 and 12 One of the SNPs those for chromosome 12 (rsID 2239067) was found to have resistance and susceptibility alleles associated with resistance or susceptibility to LOAD

SNPs associated with LOAD Using a genomic control-adjusted trend test p-value cutoff of 0.0005, a total of 53 of the 23,319 passing SNPs passed this cutoff. We do not precisely know the expected number of SNPs that should pass this criterion, as the pooling significantly enriched for large allele frequency differences. However, this set of SNPs is likely enriched for true positive associations.

In addition to the above-mentioned SNPs in the APOE region of chromosome 19, three of these 53 SNPs are in the amyloid beta (A4) precursor protein (APP) gene and one is in the presenilin 1 (PSENl) gene. Although mutations in both of these genes are known to cause the rare early-onset form of Alzheimer's disease (Bertram & Tanzi, Hum MoI Genet 13 Spec No 1 , Rl 35-41 (2004)), there has been no previous evidence of an association between either of these genes and the common late-onset form of Alzheimer's disease. These genes were not in the candidate regions we studied, and were instead identified by the two-stage approach in which : pooled genotyping identified SNPs likely to be associated with the trait, followed by individual genotyping to identify true associations. Significance

We have shown that our two-stage whole-genome association approach for identifying genes associated with LOAD has enough power to identify a large fraction of the SNPs with small allele-frequency differences between our case and control samples. We have identified a number of SNPs that show large differences in allele frequency between the case and control samples. Two genes near these SNPs, APP and PSENl, are particularly exciting, as they are known to be responsible for rare cases of early-onset Alzheimer's disease but not LOAD.

Example 12

Both the pooled genotyping and individual genotyping phases of this study involved amplification of samples by short-range PCR. For pooled genotyping, each pool was subjected to a plurality of multiplex (>100-plex) short-range PCR reactions using primers designed to amplify genomic DNA containing 267,852 SNPs. For individual genotyping, a sample from each individual was subjected to a plurality of multiplex (>100-plex), short- range PCRs using primers designed to amplify genomic DNA containing approximately 20,000 potentially associated SNPs that were identified in the pooled genotyping methodology as well as almost 5,000 SNPs from candidate regions thought to contain loci associated with LOAD. The PCRs were performed in 384-well plates containing DNA template (IOng) and PCR cocktail (1.47 μl 1OX AK2 buffer (0.5M Trizma, 0.14M ammonium sulfate, and 27mM MgC12), 0.03M tricine, 0.67μl MasterAmp 1OX PCR Enhancer (Epicentre, Madison, WI), 3.9% DMSO, 0.05M KCl, dNTPs (0.54 mM each), PCR primers (0.42 pmol/μl/primer), and -~2X Titanium Taq polymerase (BD Biosciences, Palo Alto, CA)). The PCR plates were sealed prior to PCR. Short-range PCR was performed for approximately three hours. The thermocycler hlock was allowed to reach 90⁰C before the PCR plates were placed in the thermocycler. The thermocycler program used for short-range PCR is identified in Table 12-1 :

Once the PCR was complete, the plates were removed from the thermocycler and were pooled as described infra. (At this point, the plates could also have been stored at -20⁰C for an extended period, if so desired.)

PCR plates containing amplified sample were spun at 1000 r.p.m. for 15 seconds in a table-top Sorvall centrifuge. Amplified samples from a single individual corresponding to a single chip (microarray) design were pooled together. The pooled samples were then arrayed into 96-well plates and quantified using PicoGreen reagent (Molecular Probes, Inc., Eugene, OR) and a SpectraFluor Tecan Plate Reader (Tecan Group Ltd., Maennedorf, Switzerland). Amplified samples that contained less than 100 ng/μl were deemed to have failed PCR and were not analyzed further.

Example 13

Post-PCR pools were subjected to treatment with shrimp alkaline phosphatase (SAP). Each treatment was performed in a well of a 96-well plate and contained 8 μg amplified sample, 5U SAP (Promega, Madison, WI), and ~1X One Phor All buffer Plus (Amersham Biosciences, Buckinghamshire, England) in a total volume of 100 μl. The reaction mixture was incubated at 37⁰C for 30 minutes, 8O⁰C for 20 minutes, and then cooled to 4⁰C. The SAP-treated samples were then labeled with biotin. (At this point, the SAP- treated sample could be stored overnight at -2O⁰C prior to biotin-labeling.)

Example 14

The SAP-treated pools were labeled with biotin. Each labeling reaction was performed in one well of a 96-well plate and contained the 100 μl volume of the SAP-treated pool plus 3 μl of 0.5mM biotin d/dd-UTP and 800U of recombinant TdT. The plate was sealed, vortexed briefly, and centrifuged at 1000 r.p.m. for 15 seconds in a table-top Sorvall centrifuge. The plate was placed in a thermocycler and incubated at 37⁰C for 90 minutes, 99⁰C for 10 minutes, and then cooled to 4⁰C. The biotin-labeled pools were hybridized to microarrays on the same day as they were labeled.

Example 15

Hybridization buffer (1.5M TMACL (tetramethylammonium chloride), 5mM Tris (pH 7.8 or 8.0), 0.005% Triton X-100, 26 pM b-948 control oligo (Genset, La Jolla, CA), and 0.05 mg/ml HS (herring sperm) DNA) was prewarmed at 60⁰C for a minimum of 30 minutes. Microarrays (e.g., chips) were prewarmed at 5O⁰C in a hybridization oven for approximately 30 minutes. 195 μl of hybridization buffer was added to each well of a 96-well plate that was prewarmed at 6O⁰C for a minimum of 30 minutes, and the plate ("hybridization plate") was sealed and returned to the heat block. The 96-well plate containing the labeled sample was centrifuged at 1000 r.p.m. for 15 seconds in a table-top Sorvall centrifuge prior to heating the plate at 99⁰C for 10 minutes and subsequently cooling the plate to 60⁰C (for no more than 5 minutes) to denature the labeled sample. Once the denaturation is complete, the denatured samples (105μl) were transferred to wells on the hybridization plate containing the 195μl aliquots of hybridization buffer, and were mixed by pipetting the solution up and down twice. The hybridization plates were resealed and returned to the 6O⁰C heat block.

The mixture containing the denatured samples and hybridization buffer was transferred to a prewarmed microarray. The array was sealed, returned to the 5O⁰C hybridization oven, and rotated at 20 r.p.m. overnight (14-19 hours). After the overnight incubation, the array was stained, washed and scanned as described below. Example 16

After incubation (i.e., hybridization), the microarray was removed from the hybridization oven and the sample was removed and stored at -2O⁰C. Then, the microarray was washed l-2x with 200 μl of IX MES/0.01% Triton X-IOO. The microarray was inverted several times to ensure that the wash solution moved freely over the surface of the microarray prior to removing the wash solution by vacuum suction.

Next, 200 μl of the "First Stain Solution" (174 μl of IX MES/0.01% Triton X-100, 25 μl of 20 mg/ml of acetylated BSA, and 1 μl of 1 mg/ml streptavidin) was added to each microarray. The microarray was inverted several times to ensure that the First Stain Solution moved freely over the surface of the microarray. Then, the microarray was rotated at 25 r.p.m. for 15 minutes at room temperature. Next, the microarray was washed with IX MES/0.01% Triton X-100 wash solution in a Perlegen RevD Fluidics Station. When the wash was finished the microarray was removed from the fluidics station and the IX MES/0.01% Triton X-100 wash solution was removed by vacuum suction.

Next, 200 μl of the "Second Stain Solution" (175 μl of IX MES/0.01% Triton X- 100, 25 μl of 20 mg/ml acetylated BSA, and 0.5 μl of 0.5 mg/ml biotinylated anti- streptavidin) was added to each microarray. The microarray was inverted several times to ensure that the Second Stain Solution moved freely over the surface of the microarray. Then, the microarray was rotated at 25 r.p.m. for 15 minutes at room temperature. Next, the microarray was washed with IX MES/0.01% Triton X-100 wash solution in a RevD Fluidics Station. When the wash was finished the microarray was removed from the fluidics station and the IX MES/0.01% Triton X-100 wash solution was removed by vacuum suction.

Then, 200 μl of the "Third Stain Solution" (174 μl of IX MES/0.01% Triton X-100, 25 μl of 20 mg/ml acetylated BSA, and 1 μl of 0.2 mg/ml streptavidin Cy-chrome) was added to each microarray. The microarray was inverted several times to ensure that the Third Stain Solution moved freely over the surface of the microarray. Then, the microarray was rotated at 25 r.p.m. for 15 minutes at room temperature. Next, the microarray was washed with IX MES/0.01% Triton X-100 wash solution in a RevD Fluidics Station. When the wash was finished the microarray was removed from the fluidics station and the IX MES/0.01% Triton X-100 wash solution was removed by vacuum suction.

Then, a wash solution of 6X SSPE/0.01% Triton X-100 was added to the microarray. The microarray was inverted several times to ensure that the 6X SSPE/0.01% Triton X-100 moved freely over the surface of the microarray before it was removed by vacuum suction. Next, a wash solution of 0.2X SSPE/0.005% Triton X-IOO that had been prewarmed to 37⁰C was added to the microarray, which was then incubated at 37⁰C for 30 minutes. The 0.2X SSPE/0.005% Triton X-IOO was removed by vacuum suction and a solution of IX MES/0.01% Triton X-100 was added to the microarray. The microarray was then inverted several times before the IX MES/0.01% Triton X-100 was removed by vacuum suction. Finally, fresh IX MES/0.01% Triton X-100 was added to the microarray, which was wrapped in foil prior to storage at 4⁰C or scanning of the microarray.

Example 17 On the same days the microarrays were stained and washed, they were scanned using an arc scanner. After scanning, the microarrays were removed from the scanner, wrapped in foil and stored at 4°C. The scan files generated by the scanner were then analyzed by software programs designed to interpret intensity data from microarrays. For the pooled genotyping, this software allowed discrimination of hybridization patterns that distinguished the case pools from the control pools. The data were analyzed according to the methods disclosed in the following U.S. patent applications, all of which are assigned to the assignee of the present applications: U.S. provisional patent application no. 60/460,329, filed on April 3, 2003, entitled "Apparatus and Methods for Analyzing and Characterizing Nucleic Acid Sequences"; and U.S. patent application no. 10/768,788, filed January 30, 2004, entitled "Apparatus and Methods for Analyzing and Characterizing Nucleic Acid Sequences".

Nucleic acids that were identified as strongly associated with the case or control group based on the pooled genotyping analysis were reanalyzed by genotyping individual samples for those potentially associated nucleic acids, as described below. As such, individual genotyping was performed on approximately 30,000 (-2%) of the original 1.7 million SNPs. For the individual genotyping, the software assigned genotypes at each SNP position for each individual in the case and control groups. The data were analyzed according to the methods disclosed in the following U.S. patent applications, all of which are assigned to the assignee of the present applications: U.S. patent application no. 10/351,973, filed January 27, 2003, entitled "Apparatus and Methods for Determining Individual Genotypes"; and U.S. patent application no. 10/786,475, filed February 24, 2004, entitled "Improvements to Analysis Methods for Individual Genotyping."

Example 18

A set of -250,000 SNPs from across the genome was used for this project. All amplified non-QC SNPs were analyzed for associations. 10 primer plates were not amplified by mistake and their SNPs were excluded from the analysis. 4 sub-pools consisting of 99 or 100 individuals for each cases and controls were designed and scanned. All sub-pools consisted of 100 individuals except for one control sub-pool that consisted of 99 individuals as one case individual did not have enough DNA to enter the pooling.

The extent of association was evaluated using 3 different tests: First, delta phat2 values were computed for all SNPs, and SNPs with more than 2 passing phat2 values for case and control pools and with standard error < 0.04 were considered to have reliable estimates of the delta phat2 and were included in selection using a cutoff on the delta phat2. Second, standard t-test p-values between the case and control phat2 values were computed. These t- test p-values were ranked for each chip design and an empirical p-value was computed as the rank of each SNP divided by the total number of SNPs analyzed.

Since the delta phat2 criterion tends to select preferentially SNPs that have large standard errors and the t-test on the other hand tends to select preferentially SNPs with small standard errors, we have also computed a corrected t-test p-values, which turns out to be a hybrid between these two methods. The t-test correction is based on the assumption that the standard error computed from the phat2 values underestimates the true standard error by not representing correctly the experimental variance. Therefore a constant is added to the standard errors and this constant is computed by minimizing the coefficient of variance of the t-test values:

Figure 18-1 : The minimum corresponded to 0.027. This value was added to all standard errors before the corrected t-test p-values were computed. The resulting corrected t-test p- values do not preferentially select SNPs from any region of standard errors (see Figure 18-2): standard error density comparison with snps selected using: black=all snps, red=t-test, blue=delta phat, green=corrected t-test

000 002 004 006 008 010 standard error

Figure 18-2 SNP selection:

SNPs for the following IG were selected from 4 different sources. First, 970 SNPs that were on the 10 unamplified primer plates on the pooled genotyping scans were selected. Second, SNPs in chromosomal regions of interest on chromosomes 10, 12 and 19 were identified and provided. In particular, 3656 SNPs were selected from a region between positions 60,000,000 and 95,000,000 on chromosome 10, 1173 SNPs were selected from a region between positions 2,000,000 and 10,000,000 on chromosome 12, and 50 SNPs were selected from a region between positions 50,000,000 and 50,200,000 on chromosome 19. Third, SNPs were selected from the PG (pooled genotyping) results using the 3 association metrics outlined above. Thus, 2930 SNPs with delta phat2 values > 0.07 (with > 2 passing phat2 values for each cases and controls and SE < 0.04), 376 SNPs with empirical p-values from uncorrected t-test < 0.0015, and 20125 SNPs with empirical p-values from corrected t- test p-values < 0.08 were selected. Due to significant overlaps between the sets (all of the snps selected using delta phat2 were also selected using the corrected t-test, and 350 SNPs selected using uncorrected t-test were also selected using the corrected t-test) the total number of SNPs selected from the PG results was 20,151 Fourth, 312 stratification SNPs were added to the SNP selection Due to some overlap between the sets, the total number of selected SNPs for IG was 25991

Individual genotyping

The custom IG (individual genotyping) chip contained 25,990 SNPs and was composed of SNPs selected from PG, representative SNPs for candidate genes on chromosomes 10, 12 and 19, 311 stratification SNPs and SNPs that failed amplification on PG chips The SNPs were categoπzed into categories by first assigning all SNPs that were selected as part of the candidate gene region to the "candidate gene regions" category (4879 SNPs), then from the remaining SNPs all SNPs that were included on the chip due to the unamphfied 10 pπmer plates in PG were assigned to "10 missing primer plates" category (970 SNPs), then again from the remaining SNPs after the above selections all SNPs that were included on the chip by selection from PG were assigned to the "pooled" category (19969 SNPs) and the remaining stratification SNPs were assigned to the "extra" category (172 SNPs) Due to a significant overlap between SNPs selected from PG and the stratification SNPs the "extra" category contains only 172 of the 311 stratification SNPs The first 3 categories of SNPs that represent SNPs included on the chip due to different reasons were analyzed separately in the round 1, the replication and also in the combined set of samples The "candidate gene regions" SNPs represent a separate hypothetical experiment and since this SNP selection was not informed by the PG their p- value significance needs to be corrected only for the size of this SNP set The "10 missing pπmer plates" SNPs were included because these SNPs were left out of PG and therefore again represent a separate hypothetical experiment that makes up for the mistake in PG and can be again analyzed separately and corrected again only for the number of SNPs in this set The "pooled" SNPs however were selected from the PG as the most likely to be associated Thus since their selection was informed by the PG their p-values need to be corrected for the number of SNPs tested in PG The Bonferroni corrections discussed here are the stπctest corrections, because due to linkage disequilibrium (LD) between SNPs it is likely that the number of independent tests is smaller than the number of SNPs However, the number of independent tests and the number of SNPs are not likely to be different by an order of magnitude, thus the correction by the true number of independent tests is not going to change the significance status of many SNPs corrected by Bonferroni, if any The LD structure between SNPs is taken into account in the computation of Familywise error rates (FWER) as discussed in the Methods section below.

The SNPs were filtered for obvious genotyping problems in round 1 and replication. SNPs that have call rate in round 1 genotyping < 80% were labeled with is_roundl_ok = 0 indicating likely problem with the round 1 genotypes. SNPs that have call rate in replication < 80 or have any of the 3 chi-square p-values measuring allele frequency differences between cases in the 3 different protocol IG analyses < 0.05/25990 or have any of the 3 similar chi- square p-value for controls < 0.05/25990 were labeled with is_replication_ok = 0 indicating likely problem with replication genotypes. The SNPs that would be labeled using the above criteria in either round 1 or replication, but were close to the top for any of the applied statistical tests were however inspected visually and labeled manually. These top 142 SNPs, witch genotypes have to be inspected visually, consist of SNPs that have FDR q- value < 50% or Bonferroni corrected p-value < 1 for any of the GC corrected tests or for any of the regression ANOVA tests described below. This filter described above and performed visually is still quite permissive, as it does not filter for Hardy- Weinberg equilibrium or other metrics. Thus this filter (is_roundl_ok = 1 and is_replication_ok = 1) is the minimal filter that needs to be applied to SNPs when inspecting their p-values and their locations in genes etc.

The 799 PG samples were individually genotyped on this custom chip as part of round 1 IG. The individual genotyping of the PG samples was done in two phases. In the first phase, we have included for each sample its best pass rate scan and a redo scan (if available) that had pass rate within 10% of the best scan pass rate. This genotyping was used to select a set of canonical scans for the samples. The best call rate scan for each sample was selected, however samples with scans that had best scan call rate < 80% were excluded. This set of 787 scans (784 NIA samples + 3 CEPH samples) was genotyped in the second phase. The second phase yielded the final genotypes for the PG samples on the IG chip.

In the replication IG round additional samples that were not part of the PG study were individually genotyped. The 966 replication samples were processed in the lab using 3 different protocols, thus these 3 sets of replication samples needed to be genotyped separately. Table 18-1 shows sample counts split by round 1 or replication, case-control and gender status:

Table 18-1

The gender assignment of samples was checked against heterozygosity on sex-linked X chromosome SNPs and against call rate on sex-linked Y-chromosome SNPs. Since there are only 4 sex-linked Y chromosome SNPs the call rate on Y is not very reliable measure. 6 replication samples with dubious reported gender were excluded from the study. The following plot (Figure 18-3) shows the distribution including the 6 wrong gender samples: sample genders red=female, black=male

o.o 0.1 0.2 0.3 0.4 0.5

X heterozygosity

Figure 18-3

Table 18-2 provides the Y chromosome call rate and the X chromosome heterozygosity of the 6 excluded samples: Table 18-2

Figure 18-4 provides Age of onset distribution of all samples:

Distribution of age of onset for Gases or hispϊtal visit age for controls.

40 60 80 100 age

Figure 18-4

Figure 18-5 provides the age of onset distribution at the extreme ages:

Distribution of age of onset for cases or hispital visit age for controls.

40 60 80 100 age

Figure 18-5

The samples were analyzed for population structure using STRUCTURE (which is available from the author's web site at <pritch.bsd.uchicago.edu>) and genotypes from the 311 stratification SNPs. The round 1 samples from PG produced some population differences, as shown in Figure 18-6:

STRUCTURE 2 cluster assignments - PG samples red=cases, white=controls

0.3 0.4 0.5 0.6 0.7 fractional ancestfy

Figure 18-6

The replication samples were more evenly distributed, as shown in Figure 18-7:

STRUCTURE 2 cluster assignments ■ replication samples red=cases, white=cøntrols

0.3 0.4 0.5 0.6 0.7 fractional ancestry

Figure 18-7

Methods

Trend score analysis:

Trend scores were computed separately for the PG samples (round 1) and replication samples as well as for the combined set. The following outlines the computation of the

Armitage's trend score % : (Δ/0²

Z¹ =

Var(Ap)

1

Kαr(Δp) = (p, + P₁₁ - 2P₁¹ J -J- + 2n_τ 2n_{C i}

Where P is the observed allele frequency difference between cases and controls, ^¹ is the overall population prevalence of the arbitrary designated "1" allele, " is the fraction of samples that have two copies of allele "1",^c andⁿτ are the number of case and control samples, respectively.

The stratified trend score was computed as follows: \s ∑lralatø

X = ∑Var(Ap)

Where the P and "^ary"P> are computed for each stratum separately and combined as noted above.

Logistic regression ANOVA test of genotype association:

The association of phenotype with the SNP genotypes was evaluated using logistic regressions and ANOVA tests. The genotype-phenotype association was computed using an ANOVA test between a simple model that includes possible covariates but does not include genotype-related terms and a model that includes these covariates, genotype and genotype- covariate interaction terms. The simple model involved gender, age at which the individual was diagnosed with Alzheimer disease or be disease free (called age from here on) and the population fractional ancestry (PFA) inferred from STRUCTURE. The second more complex model with genotype had additional terms consisting of genotype, gender-genotype and age- genotype interaction. The gender-genotype interaction term models possible differences in genotype effect between males and females and the age-genotype interaction models possible differences in genotype effect at early and late ages. Thus these two logistic regression models were fitted for each SNP:

Phenotype ~ gender + age + PFA Phenotype ~ gender + age + PFA + genotype + gendeπgenotype + age:genotype

The above model formulas are specified in S notation.

The ANOVA with chi-squared test was used to evaluate the differences between residual deviances of the first and second model with degrees of freedom corresponding to the number of additional terms fitted in the second model compared to the first model.

Genomic control correction:

The trend scores were corrected using GC correction. A set of SNPs that is unbiased in respect to the inflation of small association p-values between cases and controls was used to compute the Genomic Control (GC) variance inflation factor. The SNPs that were included in the IG either by selection from pooled SNPs (see section above) or as representative SNPs for the selected candidate gene regions, were excluded from the set that was used to evaluate the GC variance inflation factor for both the round 1 and combined sample analyses. SNPs that were selected from PG are expected to show inflation of small p-values due to either population structure or random sampling differences or true associations with phenotype between cases and controls among the PG samples. These SNPs therefore will show inflation of small p- values in IG if the samples that were used in the PG are included in the analysis and thus these SNPs were excluded from the GC variance inflation analysis of round 1 and combined sample set. The candidate gene region SNP representatives also have a non-neutral prior probability to show association between cases and controls and thus were also excluded from the GC variance inflation analysis for round 1 and combined set of samples. Thus, the unbiased set of SNPs consisting of stratification SNPs and SNPs that were added due to failure of amplification in the PG were analyzed for GC variance inflation for round 1 and combined sample set after filtering for call rate > 80% and for SNPs that do show all 3 genotypes (homozygous reference and alternate and heterozygous). The full set of SNPs was used in replication after applying the same filter. This set of "good" and unbiased SNPs was used for the analysis of the GC variance inflation factor that was either executed by regression (Devlin and Roeder Biometrics 55, 997-1004, 1999) or as a simple correction by the trend score mean (Devlin et al, Nature Genetics 36(11), 1129-1130, 2004). The regression allowed us to better distribute the GC correction between SNPs with varying reliability of the allele frequency difference estimate. The reliability of the allele frequency differences of SNPs was estimated by the absolute values of deltas between allele frequency difference between cases and controls computed from filtered and unfiltered genotypes. The larger the delta between the allele frequency differences of unfiltered versus filtered genotypes, the larger is the possible distortion of the allele frequency difference in the filtered genotypes caused by the genotype filtering. The regression of the trend score values against the deltas of the allele frequency differences was done using log link and Gamma distribution. This procedure allows us to better distribute the power hit from the GC correction between SNPs based on their reliability of the delta allele frequency between cases and controls. The GC correction by regression therefore yielded a GC correction specific to each SNP computed from its delta. An F-test was used to compute the GC corrected trend score p-values (Clayton et al, Nature Genetics , 37 (11), p 12431246, 2005).

ANOVA was applied to test for significance of the slope regression coefficient. If the slope coefficient was not significant (p>0.05) we have used GC correction by trend score mean instead [reference], where the variance inflation factor is estimated by the mean trend score and an F-test is used to compute the corrected trend score p-values.

For sex-linked SNPs the GC correction variance inflation factor ^ was corrected for the smaller number of chromosomes due to the presence of males among the samples:

K_→ = I₊ V R_y

Where:

R_x = -

^Ry^~ i I ) and where^C-^F ,^CM ,^T-^F , ™ are number of female cases, number of male cases, number of female controls and number of male controls, respectively. The "*"^■■* and ""'•* are the corrected ^ for chromosome X and chromosome Y sex-linked SNPs, respectively.

Familywise error rate:

Familywise error rate (FWER) was computed using permutations and the stratified trend scores. 1000 case - control assignment permutations were created while conserving each sample's stratum assignment. These permutations were used to compute 1000 stratified trend scores for each SNP. The FWER was computed as the number of permutations that yielded any stratified trend score higher than the SNPs' stratified trend score, divided by the total number of permutations (1000 in this case). The FWER reflects any LD structure among the SNPs, because the LD structure is preserved in permutations and because the FWER is the probability of making one or more Type 1 errors in the set of tests at different significance levels. These p-values are superior to Bonferroni corrected p-values which do correctly represent the number of independent tests. The FWER scores also represent exact p-values with no assumptions on the parametric distribution of the stratified trend scores.

Results Analysis details of PG samples (Round 1 IG):

Trend scores were computed from the genotypes of the PG samples and corrected using GC variance inflation factors (for details see Methods section). A variance inflation factor of 1.197 was computed from the mean trend test scores from the 1184 strat SNPs and SNPs added from the 10 unamplified primer plates in PG after applying a filter for SNPs with more than 80% call rate and SNPs that have all 3 genotype clusters. The variance inflation factor revealed some population structure or sampling differences between cases and controls and the STRUCTURE program confirmed some differences in population fractional ancestry distributions assuming two ancestral populations (see above). The GC regression (see Methods section for details) did yield a significant slope (p=0.002), thus the regression variance inflation factors were computed individually for each SNP. The following Q-Q plot (Figure 18-8) shows that the GC corrected trend score p- values are well distributed as expected under the null distribution for the stratification and 10 unamplified primer plate SNPs:

Q-Q plot for round 1 1G

5 10 quantiles of F distribution with df=(1 , 1180) Figure 18-8 Logistic regression ANOVA tests were computed as described in the Methods section for all SNPs. The following Q-Q plot (Figure 18-9) for autosomal SNPs with call rate > 80% and 3 genotype clusters in "10 missing primer plate" and stratification SNPs shows that the deviance is distributed as expected from the chi-squared distribution:

Q-Q plot round 1 , logistic regression

5 10 15 20 quantiles of chisq distribution with df=3

Figure 18-9

False Discovery Rate (FDR) q-values were computed separately for all the above test's p-values for SNPs in "candidate gene regions" and "10 missing primer plates" using Storey procedure.

P-values in the 3 SNP sets ("candidate gene regions", "10 missing primer plates", "pooled") were corrected separately using Bonferroni correction. The "candidate gene region" and the "10 missing primer plates" were corrected by the number of SNPs with call rate > 80% that are polymorphic. The "pooled" SNPs were corrected by the number of SNPs that were analyzed in the pooled genotyping stage (all SNPs that had > 1 passing subpools for both cases and controls).

PG power: Below are histograms of p-value distributions for the "candidate gene regions" SNPs and for SNPs selected from PG. The expected and actual number of significant SNPs among the "pooled" SNPs at different significance cutoffs demonstrates that the PG worked: Table 18-3

|GC corrected trend |exρected number of significant |observed number of significant |

The "expected number of significant SNPs" is just the number of false positives that we would expect at the given level of statistical significance (i.e. 18576*significance cutoff). It is obvious that the PG must have enriched for large case-control differences. Only SNPs that were selected from PG and have call rate > 80% were counted. There are 18,576 of such SNPs.

p-value distribution p-value distribution snps not selected from PG snps selected from PG

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

GC corrected (rend score p GC corrected trend score p Figure 18-10

The IG results were used to estimate the PG power. The power was evaluated over SNPs that were selected for the IG chip independent of the PG results (the candidate gene SNPs, and the stratification SNPs), but that were also analyzed in the PG study. Therefore the estimate gives us a true unbiased estimate of the power since the set of SNPs is an unbiased subset from the PG chips. This set of SNPs was further filtered for call rate > 80% and Hardy Weinberg Equilibrium p-value > 0.0001 for both cases and controls to ensure good genotyping and therefore an accurate estimate of the allele frequency of cases and controls in the IG. Table 18-4 gives the estimates of power (in %) to select from PG SNPs with varying allele frequency differences using different cutoffs for the corrected t-test empirical p-value:

Table 18-4

IG PG significance cutoff for corrected t-test empirical p-value

PG selecte POWER d 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2

Since approximately 8% of SNPs were selected from PG to IG, the PG significance cutoff that applies to our study design is 0 08 Analysis details of replication samples

Standard trend scores and protocol-stratified trend scores were computed and corrected using GC variance inflation factors (for details see Methods section) A variance inflation factor of 1 099 was computed from the mean trend scores and 1 086 from the protocol-stratified trend scores from all autosomal SNPs after applying a filter for call rate > 80% in all 3 protocol IG analyses, for SNPs that have all 3 genotype clusters, for all 3 chisq tests for allele frequency differences in controls between different replication protocols > 0 05/25990 and also the similar 3 chisq tests for cases > 0 05/25990 (20,015 SNPs after the filtering) The variance inflation factor revealed minimal population structure or sampling differences between cases and controls and the STRUCTURE program confirmed that the population fractional ancestry distributions are very similar tor cases and controls (see above) The GC regression (see Methods section for details) did yield a significant slope for both the standard trend score (p=0 002) and the protocol-stratified trend score (p=0 0096), thus the regression variance inflation factors were computed individually for each SNP

The following Q-Q plot (Figure 18-11) shows that the GC corrected trend score p- values are well distributed as expected under the null distribution for all SNPs except SNPs on chromosome 19 A similar Q-Q plot for the protocol-stratified trend score is provided in Figure 18-12 Q-Q plot for replication trend score

5 10 15 20 quanfiles of F distribution with df=(1 ,20008) Figure 18-11

Logistic regression ANOVA tests were computed as described in the Methods section for all SNPs. The following Q-Q plot (Figure 18-13) for autosomal SNPs with call rate > 80% in all 3 protocol IG analyses, with all 3 genotype clusters, with all 3 chisq tests for allele frequency differences in controls between different replication protocols > 0.05/25990 and also with the similar 3 chisq tests for cases > 0.05/25990 in "10 missing primer plate" and stratification SNPs shows that the deviance is distributed as expected from the chi-squared distribution:

Q-Q plot for replication stratified trend score

S 10 15 20 quarttites of F distribution with df=(1 ,20014) Figure 18-12

Q-Q plot for replication, logistic regression

5 10 15 quantiles of chisq distribution with df=3

Figure 18-13 The replication logistic regression was not stratified on the three different protocols that were genotyped separately, therefore a filter must be applied to make sure that the genotype clustering is consistent between the 3 protocols. The chi-square test for the allele frequency differences for controls between the 3 different genotyping analyses and the similar test for cases will reveal the inconsistencies. Therefore when looking at the logistic regression ANOVA p-values a filter must be applied to eliminate SNPs with any of the above mentioned six chi-square p-values < 0.05/25990. FDR q- values were computed separately for the 3 SNP sets "candidate gene regions",

"10 unamplified primer plates" and "pooled". Since the replication sample set did not inform the SNPs selected from PG, we can compute valid FDR q-values for the "pooled" SNPs. Bonferroni corrected p-values were also computed for the 3 SNP sets. Each of the SNP sets were corrected separately by their number of SNPs that have call rate > 80% and are polymorphic.

Analysis details of combined sample set:

Standard trend scores and trend scores stratified on the four IG analyses were computed. The stratification on the IG analyses ensures that the case-control imbalances in the replication set done with 3 different lab protocols are handled correctly after being genotyped separately and also that the replication and round 1 genotypes are correctly combined. Since we have seen some population structure differences between round 1 and replication, the trend score stratification corrects for these and other possible confounding differences between round 1 and replication. Both sets of trend scores were corrected using GC correction. The GC variance inflation factor, computed over all autosomal SNPs in the "10 missing primer plates" and stratification SNPs that did have all 3 genotype clusters, had call rates in all 4 IG analyses > 80% and had all 3 chisq tests for allele frequency differences in controls between different replication protocols > 0.05/25990 and also the similar 3 chisq tests for cases > 0.05/25990 (961 SNPs after the filtering). This stringent selection ensures that all SNPs that are used for the computation of GC variance inflation factor are well genotyped. The GC variance inflation factor was 1.078 for trend scores and 1.088 for the stratified trend scores. The GC regressions (see Methods section for details) did not yield statistically significant slopes (p=0.97 for trend scores and p=0.70 for stratified trend scores), due to the small number of GC SNPs and the relatively small overall variance inflation. Thus the p-values were corrected for all SNPs using the mean trend scores with a uniform GC variance inflation factor over the sets of autosomal and sex-linked SNPs (see Methods section for details).

The following Q-Q plot (Figure 18-14) demonstrates that the GC corrected trend score p-values are well distributed as expected under the null distribution for the stratification and 10 unamplified primer plate SNPs. A similar Q-Q plot showing the stratified trend score is provided in Figure 18-15.

Q-Q plot for all combined samples trend score

T5 10 C

JS

Xi (D

O O ϋ

CD

5 10 quanfiles of F distribution with df=(1 ,960) Figure 18-14

Q-Q plot for all combined samples stratified trend score

5 10 quarrtiles of F distribution with df=(1 ,960)

Figure 18-15

Logistic regression ANOVA tests were computed as described in the Methods section for all SNPs. The following Q-Q plot (Figure 18-16) for autosomal SNPs with call rate > 80% in all four IG analyses (one round 1 and three replication), with all 3 genotype clusters, with all 3 chisq tests for allele frequency differences in controls between different replication protocols > 0.05/25990 and also with the similar 3 chisq tests for cases > 0.05/25990 in "10 missing primer plate" and stratification SNPs shows that the deviance is distributed as expected from the chi-squared distribution:

Q-Q plot for all combined samples logistic regression

6 10 15 quaπtiles of chisq distribution with df=3 Figure 18-16

The logistic regression was not stratified on the three different protocols that were genotyped separately with a fourth stratum for the round 1 samples, therefore a filter must be applied to make sure that the genotype clustering is consistent between the 3 replication protocols. The chi-square test for the allele frequency differences for controls between the 3 different genotyping analyses and the similar test for cases will reveal the inconsistencies in replication. Therefore when looking at the logistic regression ANOVA p-values a filter must be applied to eliminate SNPs with any of the above mentioned six chi-square p-values < 0.05/25990.

FDR q-values were computed for the GC corrected p-values and the logistic regression ANOVA p-values for the two SNP sets separately: "candidate gene regions", "10 missing primer plates".

P-values in the 3 SNP sets ("candidate gene regions", "10 missing primer plates", "pooled") were corrected using Bonferroni correction separately for each of the SNP sets. The "candidate gene region" and the "10 missing primer plates" were corrected by the number of polymorphic SNPs with call rate > 80% in both round 1 and replication. The "pooled" SNPs were corrected by the number of SNPs that were analyzed in the pooled genotyping stage (all SNPs that had > 1 passing subpools for both cases and controls).

Candidate gene regions SNPs:

There are 4 SNPs that are significant in the "candidate gene regions" after Bonferroni correction in round 1 after applying filter for is_roundl_ok = 1 (described above in "IG,

SNPs and samples"). These SNPs are SNP ID nos: 3509530, 4181487, 4109479, and 884196. From these 4 SNPs the first 2 SNPs are also significant in replication, corrected only to the 4 tests. These 2 SNPs are about 10 kb apart on chromosome 19. The first SNP is in an exon region of TOMM40 gene and the second is downstream of PVRL2 and upstream of TOMM40. The remaining 2 SNPs that were significant in round 1 and did not replicate, are also not significant in the combined sample set for both the stratified and unstratified GC corrected trend scores and the logistic regression GC corrected ANOVA test when corrected for the number of tested SNPs from the "candidate gene region". Two additional SNPs reached significance when evaluated on the combined sample set using the stratified GC corrected trend scores: SNPJD nos 4290160 and 4587235. These two SNPs are positioned in the intron and exon of PVRL2 gene on chromosome 19 near the above hits.

There are 69 SNPs that have FDR q- values of the GC corrected stratified trend test for combined sample set < 50% after applying filter for is_roundl_ok = 1 and is_replication_ok = 1 (described above in "IG, SNPs and samples"). The following plots in Figure 18-17 show the distribution of SNPs on Chromosome 10, 12 and 19:

Distribution of SNPs on chromosome 10 with GC corrected trend score FDR q-value < 50%

βO e+O7 65 e+O7 70e+07 75 e*07 80e+07 chromosome 10 position

< 50%

chromosome 12 position

Distribution of SNPs on chromosome 19 with GC corrected trend score FDR q-value < 50%

Figure 18-17

The following is a list of genes and the number of SNPs (of the 69 selected above) that map on Build 35 to within 10kb of these genes. The first column provides the chromosome number, or "CHROMOSOME ID." Also included is the minimum and maximum r-squared between all possible SNP pairs (from the list of 69 SNPs) in each gene:

Table 18-5

The mm and max r-squared values give an indication of the amount of LD between the SNPs and therefore of the independence of their observed p-values

SNPs selected from PG [0001] The p-values of the SNPs selected only from PG computed in round 1 and the combined sample set are not directly comparable to the p-values of SNPs that were selected as part of the "candidate gene regions" without correcting the p-values for the number of tests performed in the two parts of the study Thus only the Bonferrom corrected p-values are comparable between the "candidate gene regions" SNPs and "pooled" SNPs. Also computation of FDR q- values is not feasible for the "pooled" SNPs evaluated in round 1 and the combined samples set. Thus in this section we will focus only on the "pooled SNPs and their p-values and Bonferroni corrected p- values. In this "pooled" SNP set there were 6 SNPs significant in round 1 after correcting for the number of tests performed in PG: SNP ID nos: 3509531, 4813800, 4331517, 4331518, 4181488, 4239119. From these 6 SNPs 5 of them (all except snp_id 4181488) have significant GC corrected trend scores in replication and all 6 of them are significant in the logistic regression ANOVA test. All of these 6 SNPs are in the vicinity of APOE and TOMM40 genes on chromosome 19.

An additional SNP is significant after Bonferroni correction in the combined sample analysis using GC corrected stratified trend score (SNP ID no. 4813803). This additional SNP is in the intron of APOE and upstream from APOCl and downstream from TOMM40 on chromosome 19. All of the 7 SNPs are also significant after Bonferroni correction using the logistic regression ANOVA test.

The replication samples provide an unbiased estimate of the FDR q-values for the "pooled" SNPs. Table 18-6 shows the distribution of SNPs that have FDR q-values < 50% for the stratified GC corrected trend scores:

Table 18-6

The round 1 samples were used for the selection of the "pooled" SNPs from PG, thus the p-values computed for the combined sample set will be influenced by any sampling and population structure differences between the cases and controls from round 1. Therefore the FDR q-values cannot be computed for the combined sample set for the "pooled" SNPs. Thus although many of the top SNPs for the combined sample set will be likely false positives, we have included SNP localizations for SNPs with GC corrected stratified trend scores < le-4 (see Table 18-7).

Table 18-7

SNPs covering the 10 missing primer plates in PG: In one instance, an analysis comprising Bonferroni correction for the number of tests in this "10 missing primer plates" SNP set was unable to find significantly associated SNPs in round 1. Example 19 In this Example, an additional 433 cases and 473 controls ("replication samples") were individually genotyped at the same 25,990 SNPs, and these genotypes along with the individual genotypes of the original 400 cases and 400 controls were used to compute a final set of delta P values. These additional samples come from two sets of replication samples. The first set consisted of 222 cases (non-familial LOAD patients) and 191 controls. These were clinic-based cases and controls. All controls were evaluated using a neuropsychological battery of tests. The frequency of evaluation of the controls depended on their age. Younger controls (60-70) were seen every 3 years, 70-80, every other year, and >80 every year. The second set of replication samples consisted of 211 familial LOAD cases and 282 controls, selected in the same way as were the original 400 cases and 400 controls. Table 19-1 column 1 shows chromosome location, columns 2-4 shows polymorphic site position as defined above. Column 6 (susceptibility allele) shows the nucleotide occupying a polymorphic site that associates with susceptibility to AD, and column 6 (resistance allele) shows the nucleotide occupying a polymorphic site that associates with resistance to AD. Column 6 shows a human genomic nucleotide segment of about 30 nucleotides including a polymorphic site represented in IUPAC-IUB ambiguity code. Table 19-1

Table 19-2 columns 1 and 3 provide references for a SNP position. Table 19-2, column 2 provides the chromosome containing a SNP. Columns 4, 6, and 8 are different statistical analyses used to determine significance of the delta P values. Column 8 lists p- values from the Cochran-Armitage trend test (Freidlin et al., Human Heredity 2002;53 : 146- 152). Column 4 lists p- values from the trend test that have been corrected for population stratification. Column 6 lists Chi-squared p-values. Column 7 lists Hardy- Weinberg Equilibrium chi-squared p-values computed from the allele frequencies of the controls. Column 5 lists delta P values (the difference in frequency of a reference allele between cases and controls).

Table 19-2

Table 19-3 lists shows the identity of the reference allele used in calculating the delta p values shown in Table 19-2.

Example 20 - Round Two Individual Genotyping Data Analysis

The total number of SNPs we attempted to genotype in round 2 IG is 50707. These include 311 stratification SNPs and 202 SNPs previously found to be on top of the list based on an analysis of the first round IG data on essentially the same samples The rest of the SNPs were selected based on pooled genotypmg results of ~1 6 million tested SNPs

The number of SNPs m the genotype report is 40432 (before any SNPs with bad genotype clustering were manually eliminated) Based on manual inspection of the genotype clustering of the top 160 SNPs in the trend test for all Caucasian samples, 16 SNPs with obvious bad clustering were identified These SNPs are excluded from further analysis Unless indicated otherwise, further analysis does not include any of the previously identified 202 SNPs 285 out of 311 stratification SNPs are included m the genotype report After eliminating the 202 previously identified SNPs and the 16 SNPs with bad genotype clustering, there were 40229 SNPs left for analysis

There were 1713 samples in the genotype report (868 cases and 845 controls) Among the 1713 samples 1676 (842 cases and 834 controls) are Caucasian, as indicated in sample manifests The other 37 samples (26 cases and 11 controls) are African Americans, Asians, Hispamcs, etc 760 samples (all Caucasian) were used in pooled genotypmg Out of these, 721 samples are in the genotype report (362 cases and 359 controls) There were 1676 - 721 = 955 non-pooled-genotyping (non-pg) Caucasian samples (480 cases and 475 controls) in the genotype report The total number of non-pg samples was 955 + 37 = 992

During analysis, it was noticed that when trend test statistics for the stratification SNPs were considered, the mean test statistic was 1 131 when 1676 Caucasian samples are analyzed and it was 1 148 when all 1713 samples are analyzed So it seems that including the non-Caucasian samples in the data analysis causes some small inflation of test statistic This is also seen for all SNPs when the non-pg samples are analyzed including the non-Caucasian samples or not

The number of non-Caucasian samples is small Although we could try to take care of the inflation of statistic caused by population structure, it is tricky in the context of this analysis (see more discussion about this topic in the next section) Given the small number of non-Caucasian samples and the fact that there are distinct populations even among the non- Caucasians, it is probably justifiable simply excluding them from our data analysis

The stratification SNPs are unbiased as far as variance inflation is concerned As mentioned above, the mean trend test statistic for stratification SNPs is 1 131 for all Caucasian samples Given the null hypothesis of no inflation of statistic, the 95% confidence interval for this mean is [0 835, 1 165] as obtained from simulation given the number of stratification SNPs Therefore, the 1 131 value is not large enough for us to rule out our null hypothesis that our variance inflation factor is 1 The test statistics for the trend test scores of the 40229 SNPs for the pg samples are not expected to follow the chi-squared distribution with 1 degree of freedom since most of the SNPs are selected from pooled genotyping data. The QQ plot in Figure 20-1 simply shows that the selected SNPs are indeed enriched for allele frequency differences between cases and controls in the pg samples.

721 pg samples 40229 SNPs

chisqH ) quantites Figure 20-1 : QQ plot of the trend test scores for pg sample data

The QQ plot of the trend test statistics for the non-pg Caucasian samples (Figure 20-2) indicates that there is a small inflation of test statistic, with a mean test statistic value of 1.036.

955 Caucasian non-pg samples 40229 SNPs

chisq(1 ) quantites

Figure 20-2: QQ plot of the trend test scores for Caucasian non-pg sample data

Principle component analysis was carried out as an attempt to figure out the source of the inflation of statistics. Using rigjhat values (instead of numeric genotype values to alleviate problems caused by missing genotypes) of about 5000 SNPs as a vector for each sample, it was found that only weights of the first two principle components are correlated reasonably well among SVD runs based on different disjoint SNP sets, with R > 0.5 consistently (R is the correlation coefficient). The weight vectors only correlate weakly with the phenotype with R < 0.1. When analysis of deviance was performed using the weights of one or both of the first two principle components as additional variables in logistic regression models, only a very small decrease in the variance inflation factor was achieved. Interestingly, when non-Caucasian samples were included in the trend tests for non-pg samples, the original variance inflation factor was 1.062. Principle component analysis identified two weight vectors that seem to be consistent among different SVD runs and both are highly indicative of population structure. When these two vectors were used in logistic regression models, the variance inflation factor was reduced to 1.035, which is very similar to what we have for the Caucasian non-pg samples only. The above evidence seem to indicate that the source of the about 0.035 inflation of statistic is hard to identify and it will be very difficult, if possible, to take care of it properly, i.e., without applying a overly stringent genomic control correction.

Applying genomic control to take care of the variance inflation in all the samples will require us to derive the variance inflation factor based on what we obtained from the non-pg Caucasian samples. To do this, we need to assume that whatever issue is causing the variation inflation in the non-pg samples is also the underlying issue for variation inflation in the pg samples and the effect is of the same magnitude. We do not really know whether we can make this assumption, especially since the great majority of the SNPs were selected based on the pooled genotyping results.

Therefore there seems to be no perfect way of adjusting our test statistics to make sure that the final test statistics has an inflation factor of 1. However, we can take some comfort in the fact that based on the results from the unbiased stratification SNPs, we know that variance inflation, if present, is unlikely to be of large magnitude. And secondly, applying the genomic control correction will not change any of our conclusions about significance, or the order of the most associated SNPs (see next section for results). The current decision is not to apply any correction at all to the trend test scores in this analysis.

Results: Association tests without considering gender information Association test of the pg sample data The QQ plot for the test statistics is shown in Figure 20-1. The largest test score is 29.15, which corresponds to a P value of 6.7e-8.

Association test of the Caucasian non-pg sample data

The QQ plot for the test statistics is shown in Figure 20-2. The largest test score is 18.84, which corresponds to a P value of 1.4e-5. Association test of all Caucasian sample data

The QQ plot for the test statistics is shown in Figure 20-3. This analysis does not identify any significant SNPs after Bonferroni correction for all tested SNPs.

1676 Caucasian samples 40229 SNPs

chisq(1) quantiles

Figure 20-3: QQ plot of the trend test scores for all Caucasian sample data

There was an absence of enrichment of small P values in replication (non-pg) samples for the top SNPs in the pg sample data analysis. For different numbers of top SNPs in the pg sample analysis, we checked to see how many of these SNPs have P values less than 0.5 in the non-pg samples. If a higher fraction than what would be expected based on the binomial distribution are identified, this finding would suggest that there is an enrichment of small P values in replication for the top SNPs found in the pg sample analysis, and therefore, although no single SNP is found to be significant at the genome level, it is very likely that some of the top SNPs in the pg sample analysis are likely to be truly associated with the phenotype. However, based on the plot in Figure 20-4, we see no sign of enrichment of small P values in replication for the top SNPs obtained from the pg sample analysis.

0 50 100 150

Number of top SNPs in pg sample analysis Figure 20-4

Out of the 202 previously identified SNPs described above, 7 did not meet our call rate threshold and another 7 did not meet our HWP-for-controls threshold. One more SNP was among the 16 SNPs identified to have bad genotype clustering. Therefore 187 SNPs were analyzed for all Caucasian samples. The top 9 SNPs in this list (with scores ranging between 26 and 223, corresponding P values ranging between 0 and 3.3e-7) are all from the PVRL2- TOMM40- APOE region on chromosomel9.

Results: association tests considering gender information

Gender-specific effects were examined by comparing two nested logistic regression models: a) phenotype ~ gender, and b) phenotype ~ gender + genotype + gendeπgenotype, Association test with the pg sample data

The maximum chi-squared statistic (df = 2) for all 40222 SNPs tested (7 SNPs on the Y chromosome were not tested) is 31.25, which corresponds to a P value of 1.6e-7. Association test of the Caucasian non-pg sample data The QQ plot for statistics obtained from testing the two genotype-containing terms is shown in Figure 20-5. This analysis did not identify any SNPs to be genome-wide significant after correcting for the number of SNPs analyzed.

955 Caucasian πon-pg samples, 40222 SNPs testing two genotype-containing terms

chisq(2) quartiles Figure 20-5: Testing genotype and genotype: gender terms

Association test of all Caucasian sample data

The maximum chi-squared statistic (df = 2) for all 40222 SNPs tested was 30.77, which corresponded to a P value of 2.1e-7.

Results: Haplotype analysis of the chrl9 region near to gene APOE

Haplotype trend regression tests were performed for sliding window sizes 1-15 for 61 SNPs with positions in the range between 49131974 and 51067619 in sequence_id 144272, which includes the TOMM40/APOE gene region, based on all Caucasian sample data. The goal of this analysis was to determine if haplotype analysis could provide any additional information.

The maximum F statistic values for each window size and each starting point are shown in Figure 20-6. The F statistics obtained for window sizes of 1 are very similar to the chi-squared statistics obtained from trend tests, reflecting results of single SNP association tests. With a window size of 1, the most significantly associated SNP is #32 (snp_id = 3509530, Table 20-1), which is in the TOMM40 gene. However, with haplotype window sizes greater than one, the most significantly associated positions are closer to the APOE gene. There are 11 haplotype alleles with F statistics larger than 300 (an arbitrary high cutoff value). SNP #38 (snp_id = 4813803), which is in gene APOE, is included in every one of these haplotypes (Table 20-2). There are 3 major haplotype alleles (allele frequency > 0.05) for each of these 11 sliding windows in the phasing results. These most likely correspond to the 3 known alleles for gene APOE. The alleles shown m Table 20-2 most likely correspond to the D4 allele of APOE.

Table 20-1: The F statistics from haplotype trend regression tests for widow size 1. All SNPs with an F statistic larger than 10 (an arbitrary high cutoff value) happen to be adjacent to each other, and they are all listed in this table ordered by their positions.

Figure 20-6- Maximum F statistics (df 1,1674) obtained in haplotype trend regression tests for different sliding windows Window sizes are shown in legend. Only haplotype alleles with at least 5% allele frequencies are tested. Missing values arise from situations where none of the haplotype alleles in a sliding window has a high enough allele frequency to be tested.

Example 21

In this Example, haplotype analysis was performed for the chromosomal region of chromosome 19 near the APOE gene. Haplotype trend regression tests were performed for sliding window sizes 1-15 for 61 SNPs with positions in the range between 49131974 and

"51067619 in contig NC OOOOl 9 8 (NCBI genome build Ti.1 ), which includes the

TOMM40/APOE gene region, based on all Caucasian sample data (1675 samples).

The maximum F statistic values for each haplotype sliding window are shown in Figure 21-1. The F statistics obtained for window sizes of 1 are very similar to the chi- squared statistics obtained from trend tests, reflecting results of single SNP association tests. With a window size of 1, the most significantly associated SNP is #32 (snp_id = 3509530, Table 21-1), which is in the TOMM40 gene. However, with haplotype window sizes greater than one, the most significantly associated positions are closer to the APOE gene. There are 11 haplotype alleles with F statistics larger than 300 (an arbitrary high cutoff value). SNP #38 (snp id = 4813803), which is in gene APOE, is included in every one of these haplotypes (Table 21-2). There are 3 major haplotype alleles (allele frequency > 0.05) for each of these 11 sliding windows in the phasing results. These most likely correspond to the 3 known alleles for gene APOE. The alleles shown in Table 21-2 most likely correspond to the D4 allele of APOE

Table 21-1

Table 21-1. The F statistics from haplotype trend regression tests for window size 1. All SNPs with an F statistic larger than 10 (an arbitrary high cutoff value) happen to be adjacent to each other, and they are all listed in this table ordered by their positions.

SNP Index SHP index

0 10 20 30 40 50

Figure 21-1 : Maximum F statistics (df 1,1673) obtained in haplotype trend regression tests for different sliding windows. Window sizes are shown in legend. Only haplotype alleles with at least 5% allele frequencies are tested Missmg values arise from situations where none of the haplotype alleles in a sliding window has a high enough allele frequency to be tested.

Example 22

In this Example, haplotypes were examined for three associated genes, PSEN2, APP, and HDAC4. First, the haplotype blocks were obtained with Haploview with the default settings using all 1713 samples. For each of the three genes, the haplotype block containing the most significant SNP (in pooled genotyping samples) in or around the gene is shown. Only haplotype alleles with frequencies larger than 5% are inspected here. The haplotype allele frequencies are obtained by fastPHASE based on all 1713 samples (allele frequencies obtained from Haploview are extremely similar). The SNP genotype shown in underlined bold font is the risky allele, and the genotype shown in underlined italic font is the protective allele. Additional data for this analysis is provided in the file

"haplotype_3genes_data_tab.txt" on the CD-R submitted herewith and incorporated herein by reference for all purposes.

Table 22-1 for PSEN2-667110

[85 kb haplotype block, contains SNPs 1193415, 4433926, 4446352, 3990234, 4433927, and

667110 (arbitrary SNP index #17-22) and three major haplotype alleles]

Table 22-2 for APP-42263

[16 kb haplotype block, contains SNPs 42182, 42256, and 42263 (arbitrary SNP index #24-

26) and three major haplotype alleles]

Table 22-3 for HDAC4-4010865

[2 kb haplotype block, contains SNPs 4010865, 4010866, and 4232917 (arbitrary SNP index

#15-17) and three major haplotype alleles (SNP #15-17, the first SNP is 4010865)]

The allele frequencies (af) and trend test scores were obtained for 958 npg samples and are shown in Table 22-4. "npg" samples are those that were not used in the pooled genotyping phase of the study. These samples were only used in the individual genotyping phase.

Table 22-4

Example 23

The goal of this example was to find SNPs that are associated with age of onset of Alzheimer's disease. All samples were Caucasian Alzheimer's patients. After samples with very early ages of onset (<50) were removed, samples in the lower and upper 25% of the age of onset distribution were used as cases and controls, respectively. Pooled genotyping was carried out on the 1.6 million SNPs described in Hinds, et al. ("Whole-Genome Patterns of Common DNA Variation in Three Human Populations" Science 307, 1072 (2005)) before SNPs were selected for individual genotyping.

The SNPs were selected based on pooled genotyping data analysis. Among 341442 SNPs analyzed in pooled genotyping, all but one SNP (with mapping problems) with empirical P values less than 0.02940 were tiled for IG (10036 SNPs). There are two IG chip designs. A set of stratification and sex QC SNPs are included on each chip design (2 stratification SNPs are in the list of 10036 SNPs from PG analysis). The total number of unique SNPs on the two IG chips is 10036 + 311 + 72 - 2 = 10417. There are 10097 SNPs with call rates no less than 0.8 and MAF greater than 0, excluding QC SNPs in the first chip design (chip_design_id = 28912). All these SNPs are included in association tests.

There were 357 samples with reported genotypes. All of them are among the 382 enc_y samples used in pooled genotyping. Two samples (ARC30220 and AR.C30221, both male and have early age of onset, 65 and 58, respectively) appear to be duplicates. Sample ARC30220 was excluded from analysis, so there were 356 samples left (181 with early disease onset and 175 with late disease onset). Samples with early disease onset had ages of onset between 50 and 68. For late onset, the range was 76 to 96. The distribution of age of onset is shown in Figure 23-1.

Distribution of age of onset for 356 samples

50 60 70 80 90

Age of onset Figure 23-1 Since most SNPs in the dataset were selected based on pooled genotyping analysis results on essentially the same sample set, PCA was carried out using only stratification SNP genotypes. The genotype clustering results of 299 reported stratification SNPs (reported according to filtering criteria on call rate, MAF cutoff, etc) from the second chip design were manually inspected and two more SNPs (snp_design_block_id's 16133441 and 16123127) were removed from PCA.

The fractions of variation (scaled eigenvalues) that corresponded to the top 10 principal components are shown in Figure 23-2. There seem to be no blatant outliers in eigenvalues. This suggests that prominent population structure was not detected with our stratification genotype data.

Fraction of variation for the top 10 PCs

Figure 23-2 Analysis of deviance based on the logistic regression model gives the following results:

Analysis of Deviance Table Model: binomial, link: logit Response: early_onset Terms added sequentially (first to last)

Df Deviance Resid. Df Resid. Dev P(>|Chi[

NULL 355 493.42 gender 1 0.02 354 493.40 0.89

PCl 1 6.22 353 487.18 0.01

PC2 1 0.16 352 487.02 0.69

PC3 1 0.10 351 486.92 0.75

PC4 1 0.29 350 486.63 0.59

PC5 1 1.13 349 485.50 0.29

PC6 1 0.89 348 484.60 0.34

PC7 1 0.88 347 483.72 0.35

PCS 1 0.46 346 483.26 0.50

PC9 1 0.26 345 483.00 0.61

PClO 1 0.02 344 482.99 0.90

ANOVA based on a linear model treating age of onset as a continuous variable gives the following results:

Analysis of Variance Table Response: age_of_onset

Df Sum Sq Mean Sq F value Pr(>F) gender 1 0.4935 0.4935 0.0050 0.943422

PCl 1 1016 1016 10.3846 0.001392

PC2 1 1 1 0.0134 0.907792

PC3 1 1 1 0.0081 0.928355

PC4 1 17 17 0.1753 0.675673

PCS 1 42 42 0.4340 0.510463

PC6 1 119 119 1.2114 0.271826 PC7 1 1 1 0.0061 0.937991 PC8 1 43 43 0.4434 0.505954 PC9 1 46 46 0.4727 0.492222 PClO 1 4 4 0.0430 0.835841 Residuals 344 33660 98

Both tests show that gender does not explain lots of variation in age of onset, and among the weights for the top 10 principal components, only the first vector explains a significant amount of variation in age of onset. The first principal component seems to be informative of population structure. So PCl is included in models for testing association.

Association tests were performed using logistic regression models:

A chi-squared statistic was obtained for each SNP by testing the following two nested logistic regression models: early_onset - PCl, and early_onset ~ PCl + genotype, where genotype is coded as 0, 1, or 2.

The histogram of P values of all analyzed 9724 SNPs included in IG due to pooled genotyping analysis results (PG) and of all 307 analyzed stratification SNPs are shown in Figure 23-3. There is clearly an enrichment of small P values for the PG SNPs. The mean test statistic for the 307 stratification SNPs is 1.075, which is well within the 95% confidence interval of (0.849, 1.164) obtained by simulation based on the null distribution.

00 02 04 06 08 1 0 00 02 04 06 08 1 0

P value P value

Figure 23-3 The test results for all 10097 SNPs analyzed were summarized and the most significant P value in the list was 2.17e-7, which can survive Bonferroni correction for 230k independent tests for a type I error rate of 0 05

Association tests were also performed using linear regression models

An F statistic is obtained for each SNP by testing the following two nested linear regression models age_of_onset ~ PCl, and age_of_onset ~ PCl + genotype, where genotype is coded as 0, 1, or 2

These tests are not independent of the tests carried out using logistic regression models The histogram of P values of all analyzed 9724 SNPs included in IG due to pooled genotyping analysis results (PG) and of all 307 analyzed stratification SNPs are shown m Figure 23-4 There is clearly an enrichment of small P values for the PG SNPs There is no strong sign of enrichment of small P values for the stratification SNPs

04 06 O B 1 0 00 02 04 06 Oβ 1 0 P value P value

Figure 23-4

The test results for all 10097 SNPs analyzed were summarized and the most significant P value in the list is 1 16e-7

Data files logistic_regression_test_results txt and lmear_regression_test_results txt each contain a header line and tab separated data fields The data fields are described in Table 23-1 Entries are ordered by P value in the data files

SNPJD Perlegen internal SNP identifier

STATISTIC The chi-squared (logistic_regression_test_results txt) statistic with 1 degree of freedom or the F statistic (hnearjregression test results txt) with dfl=l and df2 specified in data field DF2

DF2 The second degree of freedom for the F statistic (in linear__regression_testj-esults.txt only) PVAL P value calculated from the chi-squared or F statistic

NEARBY_GENES The genie environment of the SNP according to NCBI Build 36.2. More information about the nearby_genes annotation can be found at pipetools/wild/perspective.aspx?action=view&page=docs:SnpGeneAnnot ations (anno.nearby_genes)

PG 1 indicates that SNP is selected based on pooled genotyping analysis results, 0 otherwise

STRAT 1 indicates that SNP belongs to the 311 stratification SNP set, 0 otherwise

SEX 1 indicates that SNP belongs to the 72 X-linked QC SNP set, 0 otherwise

Table 23-1

Discussion: The D4 allele of the APOE gene (accession NC_000019.8, position 50100879 to 50104490 according to NCBI Build 36.2) is known to be associated with early age of onset for Alzheimer's. Among the SNPs analyzed in pooled genotyping, there are 34 SNPs within 50kb of the APOE gene. Five of them have empirical P values less than the cutoff (0.02940) for being included in individual genotyping (Table 23-2).

Table 23-2

In individual genotyping, the clustering result for SNP 4239119 is not good. Therefore that SNP is not included in the current analysis. The other 4 SNPs (4331518, 4181487, 4813593 and 4331517) are ranked number 275, 480, 389, and 721, respectively, in logistic regression test results, and number 309, 909, 561, and 846, respectively, in linear regression test results according to significance. The smallest P value in these test results is 0.0018 for SNP 4331518 in the logistic regression test results.

Example 24

Individual genotyping data for 1713 samples is provided in the file entitled "data.txt," which is on the CD-R filed herewith and incorporated herein by reference. The following analyses were performed on this individual genotyping data. The "use_for_case_control_analysis" column specifies whether a sample is included in the analysis. If a sample is labeled as Caucasian and is not one of the 3 excluded samples (90C02115, 90C03354, and CBT82001), the use_for_case_control_analysis value is 1. There are 1675 such samples. The "use_for_age_of onset analysis" column specifies whether a sample is included in the current age of onset analysis. If a sample is a case, and it has a use_for_case_control_analysis value of 1, and it is not sample ARC30679 (with inconsistent age of onset info), then the use for age_of_onset value is 1. There are 835 such samples. The "g[#]" columns are genotypes for the samples. "0" means homozygous reference; "1" means heterozygous, and "2" means homozygous alternate. Missing values are possible, gl is a haplotype allele genotype, and is the best estimate of the allele genotype obtained by the fastPHASE software. All other genotypes are for single SNPs. g2-g9 are the genotypes of all single SNPs for F test statistics of larger than 50. They are all SNPs in the TOMM40/APOE region with individual genotype data in the study comprising pooled genotyping of ~1.6 million SNPs. Tests were carried out to further test the association between late-onset Alzheimer's disease and the three SNPs in each of PSEN2, APP, and HDAC4, in part by excluding early- onset cases from the analysis, and examples of the results are found on Tables 24-1, 24-2, 24- 3, 24-4, and 24-5. The 717 "pg" samples include 358 cases and 359 controls, and the 958 "npg" samples include 478 cases and 480 controls. Table 24-1

In the analysis shown in Table 24-2, the 683 "pg" samples with age-of-onset below 65 include 324 cases and 359 controls, and the 832 "npg" samples with age-of-onset above 65 include 370 cases and 462 controls.

Table 24-2

In the analysis shown in Table 24-3, the 504 "pg" samples with age-of-onset below 70 include 225 cases and 279 controls, and the 625 "npg" samples with age-of-onset above 70 include 255 cases and 370 controls.

Table 24-3

IN THE ANALYSIS SHOWN BELOW. THE GENOTYPES OF SNPS (IDENTIFIED BY SNP ID NUMBER) FOR THREE GENES (T>SEN2, APP, AND HDAC41 ARE ANALYZED IN 717 PG CAUCASIAN SAMPLES AND TABLE 24-4 PROVIDES A COUNT OF INDIVIDUALS WITH THE GIVEN GENOTYPE WITHIN CASES AND CONTROLS. AND PROVIDES THE BASIS OF DETERMINING THE "RISKY ALLELE." (I.E.. THE ALLELE CONFERRING THE GREATER RISK OF AD-RELATED DISEASEV

TABLE 24-4

LIKE TABLE 24-4. TABLE 24-5 PROVIDES ANALYSIS OF THE GENOTYPES OF THE SAME SNPS FOR THE THREE GENES (PSEN2. APP. AND HDAC4). THIS ANALYSIS WAS PERFORMED USING GENOTYPES FROM 958 PG CAUCASIAN SAMPLES AND THE ALLELES CONFERRING THE GREATER RISK OF AD- RELATED DISEASE WERE IDENTIFIED AS THE SAME ONES IDENTIFIED IN TABLE 24-4. THUS. THE DIRECTION OF EFFECT FOR THESE SNPS ARE THE SAME IN THE NPG AND PG SAMPLES. ADDITIONALLY. THE ODDS RATIO CONFERRED BY EACH ADDITIONAL RISKY ALLELE IS GIVEN. THIS ODDS

RATIO IS AN UNBIASED ESTIMATE OF THE CAUCASIAN POPULATION EFFECT SIZE BECAUSE THESE SAMPLES WERE NOT PART OF THE DISCOVERY PHASE.

TABLE 24-5

ANOTHER SET OF ANALYSES ARE SHOWN IN THE TABLE PROVIDED IN THE FILE "DATA DETAIL-ALLELE FREO.TXT" ON THE CD-R SUBMITTED HEREWITH AND INCORPORATED HEREIN BY REFERENCE. TABLE 24-6 PROVIDES THE R2 OF THE 12 GENOTYPES. G1-G12. G1-G9 ARE

ALL FROM THE APOE REGION AND ARE IN LINKAGE DISEQUILIBRIUM TO ONE ANOTHER TO SOME DEGREE. GlO. Gl 1. AND Gl 2 (IN PSEN2, APP, AND HDAC4. RESPECTIVELY) ARE NOT IN LINKAGE DISEQUILIBRIUM WITH EACH OTHER OR WITH G1-G9. G3. G4. G5. G6. G7. AND G9 ARE ALL IN STRONG LINKAGE DISEQUILIBRIUM WITH EACH OTHER. G8. WHICH CORRESPONDS TO THE SNP PRESENT IN ALL THE MOST SIGNIFICANT HAPLOTYPE ALLELES. IS NOT IN STRONG LINKAGE DISEQUILIBRIUM WITH ANY OF THE OTHER APOE REGION SNPS. THIS SNP IS ALSO THE LEAST SIGNIFICANTLY ASSOCIATED TO THE CASE-CONTROL STATUS COMPARED WITH THE OTHER APOE ALLELES LISTED HERE.

TABLE 24-6

Table 24-7 provides analyses for testing an association between genotype and age-of- onset. Only data for 835 case samples were used for the analysis. The association was tested using linear models. The genotype was treated either as a numeric variable (Test 1) (i.e., assuming linear effects on the age-of-onset), or as a factor (Test 2) (allowing the 3 genotypes to have 3 different effect sizes on age-of-onset). The null hypothesis was that the age-of-onset in the case samples was independent of the genotype.

Table 24-7 effect size standard

P value for risky error for P value

(test 1) allele effect size (test 2) gi 3.8E-08 -2.20 0 40 2.7E-07 g2 1.2E-04 -1.54 0.40 4.6E-04 g3 1.6E-03 -1.35 0.43 2.4E-03 g4 1.1E-02 -1.18 0.46 1.9E-03 g5 2.8E-04 -1.52 0.42 6.1E-04 g6 4.9E-04 -1.46 0.42 1.0E-03 g7 1.4E-03 -1.36 0.42 2.4E-03 g8 1.8E-02 -1.07 0.45 1.7E-02 g9 1.4E-04 -1.67 0.44 7.2E-04 glO 0.998 0.00 0.46 0.144 gl l 0.614 -0.24 0.48 0.880 gl2 0.667 0.23 0.52 0.463

The two tests gave consistent results. All the APOE alleles have significant effects on age-of-onset in the 835 case samples. For gl (the 3-SNP haplotype allele), the age-of-onset decreases by about 2.2 years for each additional risky allele in the genotype. (The mean ages- of-onset are 69.42, 71.56, and 73.78, respectively, for different gl genotypes in our data).

It is to be understood that the above description is intended to be illustrative and not restrictive. It readily should be apparent to one skilled in the art that various embodiments and modifications may be made to the invention disclosed in this application without departing from the scope and spirit of the invention. The scope of the invention should, therefore, be determined not with reference to the above description, but should instead be determined with reference to the appended claims, along with the foil scope of equivalents to which such claims are entitled.

Claims

WHAT IS CLAIMED IS:

1. A method of polymorphic profiling an individual comprising: determining a polymorphic profile in at least two but no more than 1000 different genomic regions, at least two of the genomic regions proximal to or including at least a portion of a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D.

2. The method of claim 1, wherein the polymorphic profile in the at least two haplotype blocks is determined at polymorphic sites in or within 10 kb of the at least two genes selected from the group.

3. The method of claim 1 , wherein the polymorphic profile is determined in at least two genes selected from the group.

4. The method of claim 1, further comprising determining the total number of resistance and susceptibility alleles in the polymorphic profile, whereby the ratio of susceptibility alleles to resistance alleles provides an indication of whether the individual has or is at risk of AD- related disease.

5. The method of claim 1, wherein the polymorphic profile is determined in an individual having a symptom of, or known susceptibility to, Alzheimer's disease.

6. The method of claim 1, wherein the at least two haplotype blocks do not include APOE.

7. The method of claim 1, wherein the method determines the polymorphic profile in at least ten haplotype blocks, each including a different gene selected from the group.

8. The method of claim 1 , further comprising selecting a treatment or prophylactic regime for an AD-related disease based on the polymorphic profile.

9. A method of diagnosing or prognosticating AD-related disease in a subject, comprising: determining a polymorphic profile in a haplotype block of a subject including gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D.

10. A method of diagnosing or prognosticating an AD-related disease in a patient, comprising determining presence of at least one susceptibility allele shown in Table E or in linkage disequilibrium therewith, the presence of the susceptibility allele indicating presence or susceptibility to the AD-related disease.

11. The method of claim 10, wherein the method determines presence of at least one susceptibility allele shown in Table E.

12. The method of claim 10, provided the determining determines at least one susceptibility allele not in or within 40 kb of a APOE.

13. The method of claim 1, 9, or 10, further comprising administering a regime effective to treat or effect prophylaxis of an AD-related disease.

14. The method of claim 15, further provided the determining determines at least one susceptibility allele not in or within 40 kb of at least one of APOE, APOCl, PVRL2, TOMM40, CLPTMl, APOC2, or APOC4.

15. The method of claim 15, wherein the determining determines presence of at least 5 susceptibility alleles in at least five different genes selected from the group consisting of those identified in at least one of Tables A, B, C, and D.

16. A method of expression profiling, comprising determining expression levels of at least 2 and no more than 10,000 genes in a subject, wherein at least two of the genes are selected from the group consisting of those identified in at least one of Tables A, B, C, and D, the expression levels forming an expression profile.

17. A transgenic non-human animal comprising a genome comprising a transgene comprising an exogenous nucleic acid encoding the protein of a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D, whereby the animal expresses the gene, and is disposed to develop at least one sign or symptom of an AD-related disease.

18. A transgenic non-human animal comprising a genome having an enhanced, inhibited or disrupted endogenous gene that is the cognate form of a human gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D, whereby the transgenic-nonhuman animal develops at least one sign or symptom of an AD-related disease.

19. A method for identifying an agent for use in diagnosis, prognosis, prophylaxis, or treatment, of an AD-related disease, comprising: contacting a polypeptide encoded by a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D, or a nucleic acid encoding the polypeptide, with an agent to be tested; assessing a level of binding of the agent to the polypeptide or a level of modulation of activity or expression of the polypeptide by the agent; and comparing the level of binding activity or expression of the polypeptide with a control sample in an absence of the agent, wherein a difference in level of binding, activity or expression in the presence of the agent relative to the control sample is an indication that the agent has activity useful in diagnosis, prognosis, prophylaxis, or treatment, an AD-related disease.

20. A method of effecting treatment or prophylaxis of an AD-related disease, comprising administering to a patient an effective amount of an agent that modulates the activity or expression of a protein encoded by a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D.

21. A computer-implemented method of identifying a polymorphic profile characterizing a patient as amenable to treatment with an agent: providing data for a first population of patients with an AD-related disease treated with the agent and a second population of patients with the disease treated with a placebo, the data comprising whether the patient reached a desired endpoint, and a polymorphic profile of the patients in the first and second populations in at least one polymorphic site in a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D; selecting first and second subpopulations from the first and second populations based on similarity of the polymorphic profile; comparing the percentage of patients in the first subpopulation reaching the desired endpoint with the percentage of patients in the second subpopulation, a significant different indicating that the polymorphic profile of the subpopulations characterizes a patient as amenable to treatment.

22. A method of screening an agent for activity in treating an AD-related disease comprising performing a primary screen to determine whether the agent affects level of expression or function of a protein encoded by a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D, and performing a secondary screen to determine whether the agent affects the AD-related disease in an animal.

23. A method of excluding an individual from a clinical trial to test a drug for treatment or prophylaxis of Alzheimer's disease, comprising determining a polymorphic profile in an individual presenting symptoms resembling Alzheimer's disease in or within 10 kb of a plurality of genes selected from the group consisting of those identified in at least one of Tables A, B, C, and D; and determining the total number of resistance and susceptibility alleles at each locus in the polymorphic profile, wherein a high ratio of resistance to susceptibility alleles is an indication the individual should be excluded from the clinical trial.

24. A method of polymorphic profiling an individual comprising: determining a polymorphic profile in at least two but no more than 1000 different haplotype blocks, at least two of the haplotype blocks including a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D.

25. A method of expression profiling, comprising determining expression levels of at least 2 and no more than 10,000 genes in a subject, wherein at least two of the genes are shown in

Table A, B, C, or D, the expression levels forming an expression profile.

26. The method of claim 25, further comprising comparing the expression levels of the genes in the subject with expression levels of the genes in a control subject known to have an AD-related disease and/or a control subject known to lack an AD-related disease, wherein similarity of expression profiles in the subject and the control subject having the AD-related disease is an indication the subject has the AD-related disease, and similarity of the expression profiles in the subject and the control subject not having the AD-related disease is an indication the subject lacks presence or susceptibility to the AD-related disease.

27. The method of claim 85, wherein the expression levels of at least two genes selected from the group consisting of those identified in at least one of Tables A, B, C, and D are determined.

28. A transgenic non-human animal comprising a genome comprising a transgene comprising an exogenous nucleic acid encoding the protein of a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D, whereby the animal expresses the gene, and is disposed to develop at least one sign or symptom of an AD-related disease.

29. A transgenic non-human animal comprising a genome having an enhanced, inhibited or disrupted endogenous gene that is the cognate form of a human gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D, whereby the transgenic-nonhuman animal develops at least one sign or symptom of an AD-related disease.

30. A method for producing a transgenic knock-out non-human animal, comprising: providing a targeting construct containing a disrupted segment of a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D; and homologously recombining the targeting construct with the genome of a cell of the animal, whereby the construct is stably integrated into the genome of the cell; and propagating a transgenic animal from the cell.

31. A method for producing a transgenic non-human animal, comprising: introducing a construct encoding and capable of expressing the protein encoded by a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D into a cell, and propagating a transgenic animal from the cell.

32. A method for identifying an agent for use in diagnosis, prognosis, prophylaxis, or treatment, of an AD-related disease, comprising: contacting a polypeptide encoded by a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D, or a nucleic acid encoding the polypeptide, with an agent to be tested; assessing a level of binding of the agent to the polypeptide or a level of modulation of activity or expression of the polypeptide by the agent; and comparing the level of binding activity or expression of the polypeptide with a control sample in an absence of the agent, wherein a difference in level of binding, activity or expression in the presence of the agent relative to the control sample is an indication that the agent has activity useful in diagnosis, prognosis, prophylaxis, or treatment, an AD-related disease.

33. A method of effecting treatment or prophylaxis of an AD-related disease, comprising administering to the subject an effective amount of an agent that modulates the activity or expression of a protein encoded by a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D.

34. A computer-implemented method of identifying a polymorphic profile characterizing a patient as amenable to treatment with an agent: providing data for a first population of patients with an AD-related disease treated with the agent and a second population of patients with the disease treated with a placebo, the data comprising whether the patient reached a desired endpoint, and a polymorphic profile of the patients in the first and second populations in at least one polymorphic site in a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D; selecting first and second subpopulations from the first and second populations based on similarity of the polymorphic profile; comparing the percentage of patients in the first subpopulation reaching the desired endpoint with the percentage of patients in the second subpopulation, a significant different indicating that the polymorphic profile of the subpopulations characterizes a patient as amenable to treatment.

35. An isolated protein encoded by a gene shown in Table A, B, C, or D.

36. The isolated protein of claim 35, wherein at least one amino acid of the gene is encoded by a codon that includes a variant form of a polymorphic site shown in Table E.

36. An antibody that specifically binds to a protein encoded by a gene selected from the group consisting of those identified in at least one of Tables A, B, C, and D.

37. A method of screening an agent for activity in treating an AD-related disease comprising performing a primary screen to determine whether the agent affects level of expression or function of a protein encoded by a gene in Table A, B, C, or D, and performing a secondary screen to determine whether the agent affects the AD-related disease in an animal.

38. A method of screening an agent for activity in treating an AD-related disease comprising exposing a transgenic animal as defined in any of claims 17, 18, and 28-31 to the agent; and determining whether the agent treats or inhibits further development of the disease in the animal model.

39. A method for identifying a polymorphic site correlated with AD-related disease or susceptibility thereto, comprising identifying a polymorphic site within a protein encoded by a gene in Table A, B, C, or D, and determining whether a variant polymorphic form occupying the site is associated with the disease or susceptibility thereto.