CA2406383A1 - A.beta.42 lowering agents - Google Patents

Abstract

The invention provides a method of preventing, delaying, or reversing the progression of Alzheimer's disease by administering an A.beta.42 lowering agent to a mammal under conditions in which levels of A.beta.42 are selectively reduced, levels of A.beta.38 are increased, and levels of A.beta.40 are unchanged. The invention provides methods and materials for developing and identifying A.beta.42 lowering agents. In addition, the invention provides methods for identifying agents that increase the risk of developing, or hasten progression of, Alzheimer's disease. The invention also provides compositions of A.beta.42 lowering agents and antioxidants, A.beta.42 lowering agents and non-selective secretase inhibitors, as well as A.beta.42 lowering agents and acetylcholinesterase inhibitors. The invention also provides kits containing A.beta.42 lowering agents, antioxidants, non-selective secretase inhibitors, and/or acetylcholinesterase inhibitors as well as instructions related to dose regimens for A.beta.42 lowering agents, antioxidants, non-selective secretase inhibitors, and acetylcholinesterase inhibitors.

Description

A(34a LOWERING AGENTS
Statement as to Federally Sponsored Research Funding for the work described herein was provided, in part, by the federal government, which may have certain rights in the invention.
BACKGROUND
1. Technical Field The invention relates to the use of A(34a lowering agents to prevent, delay, or reverse the progression of Alzheimer's disease. The invention also relates to methods and materials involved in identifying AJ34a lowering agents that can be used to prevent, delay, or reverse Alzheimer's disease as well as methods and materials involved in identifying agents that (1) increase the risk of developing or (2) hasten the progression of Alzheimer's disease in a mammal.
~. Background Information Alzheimer's disease (AD) is the most common form of age-related neurodegenerative illness. The defining pathological hallmarks of AD are the presence of neurofibrillary tangles and senile plaques in the brain. Amyloid (3 polypeptides (A(3) are the major constituents of amyloid plaques and are derived from altered processing of amyloid precursor proteins (APPs). A(3 consists predominantly of two forms, A(34o and A(34a. Although A[34o is the predominant form, recent evidence suggests that A(3421S the pathogenic form. In addition to A(34o and A(34a, the processing of APP
generates other A[3 forms such as A~39, A[338, A(33~, and A(334.
Genetic predisposition is the largest cause of AD in the population, accounting for perhaps 50% or more cases of this disorder (Blacker et al. (1990 Arch Neurol 55:294-6).
In the past decade, epidemiological evidence suggests that non-steroidal anti-inflammatory drug (NSAm) treatment, estrogen replacement therapy, and antioxidant therapy may have beneficial effects in AD. Experimental support for these treatment methods, however, is indirect. In addition, there is no convincing evidence from randomized clinical trials that any medication tested to date slows the progression of AD.
The rational development of compounds that influence key pathways or targets involved in the development of AD is critically important.
SUMMARY
The invention relates to the use of A/342 lowering agents to prevent, delay, or reverse the progression of AD. The invention is based on the discovery that some but not all NSAms useful for treating AD are those that can selectively reduce the level of the pathogenic A(342 form, do not affect the level of A(34o, and increase levels of A(3 forms smaller than A(34o such as A(338. More specifically, the invention provides methods and materials related to identifying A(342 lowering agents, including NSAIDs, NSAID
derivatives, and NSAID analogues, that (1) can reduce the level of Aj342 by reducing APP
processing into A(342 or by increasing AJ34a catabolism; (2) increase the level of Aj33$ by increasing APP processing into A(338; and (3) have increased selectivity for reduction of AJ342 relative to inhibition of COX-1, COX-2, or both COX-1 and COX-2. In addition, the invention provides methods and materials related to identifying agents that can increase the risk of developing AD, or hasten the progression of AD, in a mammal. The invention also provides compositions and kits that can be used to prevent, delay, or reverse the progression of AD.
In one embodiment, the invention provides a method of preventing, delaying, or reversing the progression of AD by administering an A(34a lowering agent to a mammal under conditions in which levels of A(34z are reduced, levels of A(338 axe increased, and levels of A~i4o are unchanged. The A(342 lowering agent also can increase the levels of one or more of A(334, A(336, A(33~, and A(339.
The A(342 lowering agent can be an NSAID, an NSAID derivative, an NSAID
analogue, or any compound that reduces levels of A(342, increases levels of A(338, and has no effects on levels of A~34o, (i.e., levels of A(34o are neither increased nor decreased). The A(342 lowering agent can be an aryl propioiuc acid derivative, an aryl acetic acid derivative, or an amino carboxylic acid derivative. More specifically, the A(342 lowering agent can be a structural derivative of an NSAID such as flufenrnic acid, meclofenamic acid, fenoprofen, carprofen, ibuprofen, ketoprofen, and flurbiprofen. The A(34a lowering agent also can be a structural derivative of 5-vitro-2-(3-phenylpropylamino) benzoic acid). Typically, the A(342 lowering agent either (1) lacks COX-1, COX-2, or both COX-1 and COX-2 inhibiting activity, or (2) has a greater potency for lowering Aj34a levels than for inhibiting COX-l, COX-2, or both COX-1 and COX-2 activity.
AJ342 lowering agents of the invention can be used to treat AD in a mammal such as a human. The mammal may not be diagnosed with AD, or may not have a genetic predisposition for AD.
In another embodiment, the invention provides a method for developing an A(34z lowering agent. The method involves generating derivatives of the NSAms meclofenamic acid or flufenamic acid by altering the position of the carboxylic acid group on the phenyl ring or altering the position or type of substituents on the phenyl ring opposite the carboxylic acid group. Derivatives also can be generated by altering the bond connecting the two phenyl rings, altering the carboxylic acid group to propionic acid or another substituent, or performing any combination of these alterations.
The derivative is then tested to determine its ability to decrease A(342 levels while increasing A~33g levels.
In another embodiment, the invention provides a method for developing an A(34a lowering agent. The method involves generating derivatives of the NSAms fenoprofen, flurbiprofen, or carprofen. Derivatives can be generated by altering the position of the propionic acid group on the phenyl ring, or altering the position or type of substituents on the phenyl ring opposite the propionic acid group. Derivatives also can be generated by altering the bond connecting the two phenyl rings, altering the acetic acid group to carboxylic acid or another substituent, or performing any combination of these alterations. The derivative is then tested to determine its ability to decrease A(342 levels while increasing A(338 levels.
In another embodiment, the invention provides a method for developing an A(342 lowering agent. The method involves generating derivatives of indomethacin by altering the carboxylic acid group to another substituent, altering the indole nitrogen to another substituent, or performing any combination of these alterations. The derivative is then tested to determine its ability to decrease A(342 levels while increasing A~i3$ levels.
In another embodiment, the invention provides a method for developing an A(34a lowering agent. The method involves generating derivatives of sulindac sulfide by altering the methylthiol group, the propionic acid group, or the fluoride moiety to another substituent, or performing any combination of these alterations. The derivative is then tested to determine its ability to decrease AJ34z levels while increasing A~i38 levels.
In another embodiment, the invention provides a method for identifying an A(34z lowering agent useful for preventing, delaying, or reversing the progression of Alzheimer's disease. The method involves treating a biological composition that has APP and an APP processing activity with a candidate A[34z lowering agent under conditions in which APP processing occurs. An A~i4z lowering agent, useful for preventing, delaying, or reversing the progression of Alzheimer's disease, is one that, when present, decreases the level of AJ34z in the biological composition.
In another embodiment, the invention provides a method for identifying an A(34z lowering agent useful for preventing, delaying, or reversing the progression of Alzheimer's disease. The method involves treating a biological composition that has A(34z and an A(34z catabolic activity with a candidate A(34z lowering agent under conditions in which A(34z catabolism occurs. An A(34z lowering agent, useful for preventing, delaying, or reversing the progression of Alzheimer's disease, is one that, when present, decreases the level of Aj34z in a biological composition.
In another embodiment, the invention provides a method for identifying an A[34z lowering agent that has a greater potency for lowering A(34z levels than for inhibiting COX-1, COX-2, or both COX-1 and COX-2 activity. The method involves identifying A(34z Lowering agents by screening for those having the ability to lower the level of A(34z in a biological composition. The ICSO of the A(34z lowering agent for A(34z lowering can be determined by performing dose response studies. The A(34z lowering agent is examined for the ability to inhibit COX-1, COX-2, or both COX-1 and COX-2 using in vitro COX-1 and COX-2 inactivation assays. The IC50 for A(34z lowering is compared to the IC50 for COX-1, COX-2, or both COX-l and COX-2 inhibition. An A~34z lowering agent that has an greater potency for lowering A(34z levels than for inhibiting COX-l, COX-2, or both COX-1 and COX-2 activity is one that has an IC50 for A~34z lowering greater than ten-fold the IC50 for COX-1, COX-2, or both COX-1 and COX-2 inhibition.
The greater potency for lowering A[34z levels than for inhibiting COX-1, COX-2, or both COX-1 and COX-2 activity can be confirmed by demonstrating that administration of the A~i4z lowering agent to an animal reduces A[34z levels at doses that do not inhibit or only minimally inhibit COX-1, COX-2, or both COX-1 and COX-2 activity such that significant COX-related side-effects do not occur.
In another embodiment, the invention provides a method for identifying an agent $ that increases the risk of developing, or hastens progression of, AD in a patient. The method involves exposing a biological composition that has APP and an APP
processing activity to a candidate agent under conditions in which APP processing occurs.
The level of A(34z in the biological composition exposed to the candidate agent is compared to the level of A(34z in a biological composition not exposed to the candidate agent.
The candidate agent is one that can increase the risk of developing, or hasten the progression of, AD if an increase in the level of A(34z in the biological composition exposed to the agent is observed when compared with the level of A~i4z in the biological composition not exposed to the agent.
In another embodiment, the invention provides a method for identifying an agent 1 S that increases the risk of developing, or hastens progression of, AD in a patient. The method involves exposing a biological composition that has A(34z and an A(34z catabolic activity to a candidate agent under conditions in which A(34z catabolism occurs. The level of A(34z in the biological composition exposed to the candidate agent is compared to the level of A(34z in a biological composition not exposed to the candidate agent.
The candidate agent is one that can increase the risk of developing, or hasten the progression of, AD if an increase in the level of A(34z in the biological composition exposed to the agent is observed when compared with the level of A(34z in the biological composition not exposed to the agent.
In another embodiment, the invention provides a composition consisting of an 2S A[34z lowering agent and an antioxidant. The antioxidant can be, without limitation, vitamin E, vitamin C, curcumin, and Gingko biloba.
Tn another embodiment, the invention provides a composition consisting of an A(34z lowering agent and a non-selective secretase inhibitor.
In another embodiment, the invention provides a composition consisting of an A(34z lowering agent and an acetylcholinesterase inhibitor.
Tn another embodiment, the invention provides kits containing (1) an A~i4z lowering agent and an antioxidant; (2) an A(34a lowering agent and a non-selective secretase inhibitor; or (3) an A(342 lowering agent and an acetylcholinesterase inhibitor.
Fits can include instructions that indicate dose regimens for the A(34a lowering agent, the antioxidant, the secretase inhibitor, andlor the acetylcholinesterase inhibitor.
In another embodiment, the invention provides for the use of an A(34a lowering agent in the manufacture of a medicament for the treatment of AD. When administered to a patient, the medicament containing the A(342 lowering agent is effective for reducing A[342 levels without affecting A(34o levels. The medicament also can increase A(338 levels, and may also increase A[334, A(336, A(33~, or A(339 levels. The A[34a lowering agent in the medicament can be an aryl propionic acid derivative, an aryl acetic acid derivative, or an amino carboxylic acid derivative. More specifically, the A[342 lowering agent in the medicament can be a structural derivative of an NSAID selected from the group consisting of flufenmic acid, meclofenamic acid, fenoprofen, carprofen, ibuprofen, ketoprofen, and flurbiprofen. The A~iaa lowering agent also can be a structural derivative of 5-nitro-2-(3-phenylpropylamino)benzoic acid). The A(342 lowering agent in the medicament can lack COX-l, COX-2, or both COX-1 and COX-2 inhibiting activity.
The A(342 lowering agent in the medicament can have a greater potency, in vivo, for lowering A(342 levels than for inhibiting COX-1, COX-2, or both COX-1 and COX-activity. The medicament can be used to treat AD in a mammal such as a human.
The medicament can be used in a mammal that has not been diagnosed with AD, or in a mammal that does not have a genetic predisposition for AD.
The term "A(342 lowering agent" as used herein refers to an NSAID, an NSA>D
derivative, an NSA)D analogue, or any compound that (1) has the ability to reduce A~i4a levels, (2) has the ability to increase A[338 levels, and (3) has no affect on A(34o levels.
The A(342 lowering agent also can increase the levels of one of A(334, A[336, A(33~, or A[339~
The A[34a lowering agent can be a derivative of aryl propionic acid, aryl acetic acid, or amino carboxylic acid. The A[34z lowering agent can be a derivative of an NSA)D such as flufenmic acid, meclofenamic acid, fenoprofen, carprofen, ibuprofen, ketoprofen, and flurbiprofen. The A(342 lowering agent can (1) lack COX-l, COX-2, or both COX-1 and COX-2 inhibiting activity; or (2) have a much greater potency, in vivo, for lowering A[34a relative to COX-1, COX-2, or both COX-1 and COX-2 inhibiting activity.

As used herein, the terms "increase" and "decrease," refer to a change in any amount that is reproducible and significant. A reproducible and significant change is differentiated from irreproducible or insignificant experimental variations in measurements by standard statistical analysis methods including analysis that involves comparison with changes observed for control agents known to have no effects on the levels of the A(3 forms of interest. A significant change can be any amount such as a 0.5, 1, 5, 10, 20, 40 or more than 40% increase or decrease.
Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, suitable methods and materials are described below. All publications, patent applications, patents, and other references mentioned herein are incorporated by reference in their entirety. In case of conflict, the present specification, including definitions, will control. In addition, the materials, methods, and examples are illustrative only and not intended to be limiting.
Other features and advantages of the invention will be apparent from the following detailed description, and from the claims.
DESCRIPTION OF DRAWINGS
Figure 1 is a bar graph summarizing A/3421A/34o ratios and total A(3 levels determined for CHO cells expressing APP751 and PS-1 mutant M146L that had been treated with DMSO or with various concentrations of sulindac sulfide.
Figure 2 is a bar graph summarizing A(34a/A(34o ratios and total A~3 levels determined for human neuroglioma cells (HS683) expressing APP695 that had been treated with DMSO
or with various concentrations of sulindac sulfide.
Figure 3 is a bar graph summarizing A[34a/A[34o ratios and total A(3 levels determined for CHO cells expressing APP75I and PS-1 mutant MI46L that had been treated with ethanol or with various concentrations of ibuprofen, Figure 4 is a bar graph summarizing A(342/A(34o ratios and total A[3 levels determined for CHO cells expressing APP751 and PS-1 mutant M146L that had been treated with DMSO or with various concentrations of indomethacin.
Figure 5 is a bar graph summarizing A(342/A(3ao ratios and total A[3 levels determined for CHO cells expressing APP751 that had been treated with DMSO or with various concentrations of naproxen.
Figure 6 is a bar graph comparing A(342/A(34o ratios and total A(3 levels in CHO cells expressing APP751 that had been treated with ethyl acetate or various concentrations of celecoxib.
Figure 7 is a bar graph summarizing A(342/A(34o ratios and total A(3 levels determined for primary fibroblasts (from COX-1/ COX-2 double-knockout mice) expressing APP

that had been treated with DMSO or various concentrations of sulindac sulfide.
Figure ~ is two representative mass spectra of A(3 species secreted by CHO
cells expressing APP751 after treatment with DMSO or 100 p,M sulindac sulfide.
Figure 9 is a bar graph illustrating ratios of A(31_42, A(31_39~ A(~i-3s~ ~d A(31-37 to A(3mo in cells treated with DMSO or sulindac sulfide.
Figure 10 is a scattergram of A(34o and A(342 levels in the brains of Tg2576 mice after short-term NSAm treatment.
Figure 11 is a summary of the structures of indomethacin and meclofenamic acid, possible side chain modifications, and the effects of these modif canons on COX-1 and COX-2 activities.
Figure 12 is a compilation of the structures of newly synthesized biphenyl amines.

Figure 13 is a time course of A(342 reduction in CHO APP695NL,I,his cell cultures treated with meclofenamic acid.
DETAILED DESCRIPTION
The invention relates to the use of A(342 lowering agents to prevent, delay, or reverse the progression of AD. The invention is based on the discovery that some but not all NSAIDs useful for treating AD are those that can reduce the level of the pathogenic A(342 form and increase the levels of A(3 forms smaller than A(34o such as A(338.
Therefore, the invention provides methods and materials related to identifying A(34a lowering agents, including NSAIDs, NSAID derivatives, and NSAID analogues that (1) can reduce the level of A(342 by reducing APP processing into A(342 or by increasing A(34a catabolism; (2) increase the level of p~(338 by increasing APP processing into A(338; and (3) have increased selectivity for reduction of A~34a relative to inhibition of COX-1, COX-2, or both COX-1 and COX-2. In addition, the invention provides methods and materials related to identifying agents that can increase the risk of, or hasten the progression of, AD
in a mammal, by increasing the processing of APP into A(342, or decreasing the catabolism of A[34a. The invention also provides compositions and kits that can be used to prevent, delay, or reverse the progression of AD.
1. A,Q42 lowening agents A(34~ lowering agents include, without limitation, NSAIDs, NSAID derivatives, and NSAID analogues. NSAIDs can be FDA-approved NSAIDs. NSAID derivatives are compounds generated by modifying functional groups of known NSAIDs. Once modified, derivatives may or may not have the anti-inflammatory properties of the parent NSAIDs. Structural analogues of NSAIDs are compounds that are structurally similar to NSAIDs. Analogues also may not have the anti-inflammatory properties of the corresponding structurally similar NSAIDs to which they resemble.
NSAIDs are non-steroidal anti-inflammatory drugs that are distinct from steroidal drugs with anti-inflammatory properties such as corticosteroids. NSAIDs, many of which are organic acids, typically have analgesic (pain-killing), anti-inflammatory, and antipyretic (fever-reducing) properties. Some examples of NSAIDs include salicylic acid (Aspirin), ibuprofen (Motrin, Advil), naproxen (Naprosyn), sulindac (Clinoril), diclofenac (Voltaren), piroxicam (Feldene), ketoprofen (Orudis), diflunisal (Dolobid), nabumetone (Relafen), etodolac (Lodine), oxaprozin (Daypro), Meclofenamic acid (Meclofen) and indomethacin (Indocin). NSAIDs can be grouped into classes, for example, amino aryl carboxylic acid derivatives (e.g., flufenamic acid, meclofenamic acid); aryl acetic acid derivatives (e.g., indomethacin, sulindac); and aryl propionic acid derivatives (fenoprofen, ibuprofen, carprofen).
Although NSAIDs have multiple cellular effects (see Cronstein et al. (1995) Ahnu Rev Pharmacol Toxieol 35:449-62; and Amin et al. (1999) Cell Mol Life Sci 56:305-12), many act through direct inhibition of COX enzymes. COX enzymes oxidize arachidonic acids from membrane bound phospholipids to prostaglandins (see Smith et al.
(2000) Ahn Rev Biochem 69:145-82). Inhibition of COX enzymes and therefore prostaglandin synthesis is believed to underlie the analgesic and anti-inflammatory properties of aspirin and NSAIDs (see Dubois et al. (1998) FASEB J 12:1063-73). There are two isoforms of COX: COX-1 and COX-2. Although COX-1 and COX-2 catalyze the same reaction, they axe derived from two different genes. COX-1 is traditionally viewed as a constitutive or housekeeping enzyme while COX-2 is viewed as an inducible enzyme that is expressed during inflammatory circumstances. COX products, primarily prostaglandin E2, modulate classical signs of inflammation. Another COX product is thromboxane A2 that promotes platelet aggregation and vasoconstriction. Although COX is expressed in neurons, its function in the central nervous system is unclear.
Another target of NSAIDs is the peroxisome proliferator-activator receptor (PPAR) family of nuclear hormone receptors. The PPAR family consists of at least three subtypes: PPARa, PPARS, and PPARy (see Corton et al. (2000) Anhu Rev Pha~macol Toxicol 40:491-518). These receptors are thought to function as ligand-dependent activators of transcription. All three PPAR members are modulated by NSAIDs, although in different ways. For example, NSAIDs activate the activities of PPARa and PPARy but inhibit PPARS activity (see He et al. (1999) Cell 99:335-45). It is known that PPARy expression is increased in brains of AD individuals (Kitamura et al. (1999) Biochem Biophys Res Commun 254:582-6), and that PPARy agonists block A(3-stimulated secretion of proinflammatory products of microglia, including IL-1 and TNF-a (see Combs et al. (2000) JNeurosci 20:558-67). It has been suggested that the beneficial effects of NSAIDs in AD may be mediated via their activity on PPARy rather than or in addition to COX inhibition (Combs et al. (2000) JNeurosci 20:55-67). It is not lmown, however, what downstream genes are activated by PPARs, or whether they are involved in A[3 production.
S An A(342 lowering agent is any compound that has the following three properties:
(1) the ability to reduce the level of A[342 either through reducing APP
processing or increasing A(342 catabolism, (2) no effect on the level of A(34o, and (3) and the ability to increase A(338. These three properties differentiate A(342 lowering agents of the invention from other compounds having COX inhibiting activities or those that do not selectively reduce A(34a production. These three properties are referred to collectively as the Alzheimer's-A~i4a-NSAID (A(342-NSAID) footprint. In addition to having the A[342-NSAID footprint, an A(34a lowering agent of the invention can modulate the level of A(3 forms smaller than A(34o such as A(334, A(336, A[33~, and A(339.
2. Identifzcation of A/342 lowering agents useful for treating AD
A(342 lowering agents can be identified from collections of NSAIDs, NSAID
derivatives, NSAID analogues, or other compounds using the A(34a-NSAID
footprint.
Such compounds can be obtained from any commercial source. For example, NSAIDs, NSAID derivatives, and NSAID analogues can be obtained from Sigma, Biomol, Cayman Chemical, ICN, or from the web through the Chemnavigator website. Novel NSAIDs, novel NSAID derivatives, and novel NSAID analogues can be chemically synthesized using methods described in many published protocols. NSAIDs, NSAID
derivatives, and NSAID analogues can be synthesized with altered potency for their known targets such as COX-1 and COX-2. For example Kalgutkar et al. (2000) PNAS 97:925-930 have made derivatives of indomethacin and meclofenamic acid and Bayly et al (1999) Biorg ahd Med Chem Letters 9: 307 312 have made derivatives of Flurbiprofen. Indeed, because of the effort to engineer NSAZI7s so that they preferentially inhibit COX-2 rather than non-selectively inhibit COX-1 and COX-2, there are dozens of published reports documenting synthesis of novel derivatives of known NSAIDs (reviewed in Dewitt (1999) Molecular Pharmacology 55:625-631).

It is recognized that some NSAID derivatives or NSAID analogues generated can have (1) increased potency for lowering A~34a levels and (2) decreased potency for COX
inhibition. Although derivatives and analogues may no longer be considered NSAIDs since they may lack anti-inflammatory properties, A(34a lowering agents can include such NSAID derivatives and NSAID analogues.
A(342 lowering agents that have the A(342-NSAID footprint can be identified using cell free assays, ih vitro cell-based assays, and in vivo animal studies.
A~i42 lowering agents can be dissolved in any suitable vehicle for ih vitro cell culture studies or ih vivo animal or human studies. A vehicle is an inert solvent in which a compound can be dissolved for administration. It is recognized that for any given A(342 lowering agent, a vehicle suitable for in vitYO cell culture studies or ih vivo animal studies may not be the same as the vehicle used for human treatment. Some examples of suitable vehicles for cell culture or animal studies include water, dimethyl sulfoxide, ethanol, and ethyl acetate.
To identify A[342 lowering agents that reduce APP processing, a biological composition having an APP processing activity (i.e. an activity that processes APP into various A(3 forms, one of which is A(342), is incubated with APP under conditions in which APP processing occurs. To identify A(342 lowering agents that increase A~i42 catabolism, a biological composition having A(342 catabolic activity is incubated with A~34a under conditions in which A[342 catabolism occurs. Depending on the nature of the biological composition, the APP or A(3a2 substrate can be added to the biological composition, or, each or both can be.a component of the biological composition. APP
processing or A(34z catabolism is allowed to take place in the presence or absence of the candidate A(342 lowering agent. The level of A(342 generated from APP
processing or the level of A(34a remaining after the catabolic reaction, in the presence and absence of the candidate A(342 lowering agent, is determined and compared. A~i42 lowering agents useful for treating AD are those that reduce the level of A(342 either by reducing APP processing into A(34a or by enhancing A(34a catabolism and increasing A(338 production.
The biological composition having an APP processing and/or catabolic activity can be a cell-free biological sample. For example, a cell-free biological sample can be a purified or partially purified enzyme preparation; it also can be a cell lysate generated from cells able to process APP into A(342 or from cells able to catabolize A[342. Cell lysates can be prepared using known methods such as, for example, sonication or detergent-based lysis. In the case of an enzyme preparation or cell lysate, APP can be added to the biological composition having the APP processing activity, or A[342 can be added to the biological composition having A[342 catabolic activity.
In addition, the biological composition can be any mammalian cell that has an APP processing activity as well as a nucleic acid vector encoding APP.
Alternatively, the biological composition can be any mammalian cell that has A(3 catabolic activity as well as a nucleic acid vector or a viral nucleic acid-based vector containing a gene that encodes A(34~. The vector typically is an autonomously replicating molecule, a molecule that does not replicate but is transiently transfected into the mammalian cell, or a vector that is integrated into the genome of the cell. Typically, the mammalian cell is any cell that can be used for heterologous expression of the vector-encoded APP or A(342 in tissue culture. For example, the mammalian cell can be a Chinese hamster ovary (CHO) cell, a fibroblast cell, or a human neuroglioma cell. The mammalian cell also can be one that naturally produces APP and processes it into A(34a, or one that naturally produces and catabolizes A(34a.
Further, the biological composition can be an animal such as a transgenic mouse that is engineered to over-express a form of APP that then is processed into A(34a~
Alternatively, the animal can be a transgenic mouse that is engineered to over-express A(342. Animals can be, for example, rodents such as mice, rats, hamsters, and gerbils.
Animals also can be rabbits, dogs, cats, pigs, and non-human primates, for example, monkeys.
To perform an in vitro cell-free assay, a cell-free biological sample having an activity that can process APP into A(34a is incubated with the substrate APP
under conditions in which APP is processed into various A[3 forms including A(342 (see Mclendon et al. (2000) FASEB 14:2383-2386. Alternatively, a cell-free biological sample having an activity that can catabolize A(342 is incubated with the substrate A(34a under conditions in which A[34a is catabolized. To determine whether a candidate A(34a lowering agent has an effect on the processing of APP into A(34a or the catabolism of A~42a ~o reactions are compared. In one reaction, the candidate A(34a lowering agent is included in the processing or catabolic reaction, while in a second reaction, the candidate A(342 lowering agent is not included in the processing or catabolic reaction.
Levels of the different A~i forms produced in the reaction containing the candidate A(342 lowering agent are compared with levels of the different A~3 forms produced in the reaction that does not contain the candidate A(34a lowering agent.
The different A(3 forms can be detected using any standard antibody based assays such as, for example, immunoprecipitation, western hybridization, and sandwich enzyrne-linked immunosorbent assays (ELISA). Different A(3 forms also can be detected by mass spectrometry; see, for example, Wang et al. (1996) JBiol Chem 271:31894-902.
Levels of A/3 species can be quantified using known methods. For example, internal standards can be used as well as calibration curves generated by performing the assay with known amounts of standards.
Ih vitro cell-based assays can be used determine whether a candidate A(34~
lowering agent has an effect on the processing of APP into A(34a or an effect on catabolism of A(34a. Typically, cell cultures are treated with a candidate A~i4a lowering agent. Then the level of A[34a in cultures treated with a candidate A[342 lowering agent is compared with the level of A(342 in untreated cultures. For example, mammalian cells expressing APP are incubated under conditions that allow for APP expression and processing as well as A(34a secretion into the cell supernatant. The level of A(34~ in this culture is compared with the level of A(34z in a similarly incubated culture that has been treated with the candidate A[342 lowering agent. Alternatively, mammalian cells expressing A(342 are incubated under conditions that allow for A(342 catabolism. The level of A(342 in this culture is compared with the level of A~i42 in a similar culture that has been treated with the candidate A(342 lowering agent.
In vivo animal studies also can be used to identify A(34a lowering agents useful for treating AD. Typically, animals are treated with a candidate A(342 lowering agent and the levels of A(342 in plasma, CSF, and/or brain are compared between treated animals and those untreated. The candidate A(34a lowering agent can be administered to animals in various ways. For example, the candidate A(342 lowering agent can be dissolved in a suitable vehicle and administered dirsctly,using a medicine dropper or by injection. The candidate A(342 lowering agent also can be administered as a component of drinking water or feed. Levels of A(3 in plasma, cerebral spinal fluid (CSF), and brain are determined using known methods. For example, levels of A(342 can be determined using sandwich ELISA or mass spectrometry in combination with internal standards or a calibration curve. Plasma and CSF can be obtained from an animal using standard methods.
For example, plasma can be obtained from blood by centrifugation, CSF can be isolated using standard methods, and brain tissue can be obtained from sacrificed animals.
When present in an in vitro or in vivo APP processing or A(342 catabolic reaction, A[34a lowering agents reduce the level of A(34z generated by APP processing or remaining following Aj3 catabolism. For example, in an in vitro cell-free assay, the level of A(342 is reduced due to either a reduction of APP processing or an increase in A[34a catabolism in the presence the A/34~ lowering agent. In an in vitro cell culture study, a reduction in the level of A~342 secreted into the supernatant results from the effect of the A[34a lowering agent on either a reduction in processing of APP into A[34a or an increased catabolism of A[34a. Similarly, in animal studies, a reduction in the level of A(34a that can be detected in plasma, CSF, or brain is attributed to the effect of the A~i4a lowering agent on either a reduction in the processing of APP into A(342 or an increase in the catabolism of A(34~.
The level of A[342 can be reduced by a detectable amount. For example, treatment with an A[34a lowering agent leads to a 0.5, 1, 3, S, 7, 15, 20, 40, 50, or more than 50%
reduction in the level of A(342 generated by APP processing or remaining following A(34a catabolism when compared with that in the absence of the A(34a lowering agent.
Preferably, treatment with the A(342 lowering agent leads to at least a 20%
reduction in the level of A(342 generated when compared to that in the absence of A(34a lowering agent.
More preferably, treatment with an A(342 lowering agent leads to at least a 40% reduction the level of A(342 when compared to that in the absence of an A(34a lowering agent.
Typically, the A[34a lowering agent-associated reduction of A[342 levels is accompanied by an increase in the level of A(338. In contrast, no change is observed in (1) the level of A(34o generated by APP processing or A(342 catabolism in cell-free assays, (2) the level of A(34o secretion into culture supernatants in cell-based assays, or (3) the level of A~34o detected in blood plasma, CSF, or brains of animals treated with A(34a lowering agent.

A(34a lowering agents of the invention may lack COX inhibitory activity or have reduced COX-1, COX-2, or both COX-1 and COX-2 activity. COX inhibitory activity can be determined using known methods. For example, COX inhibitory activity can be determined using the method described in Kalgutkar et al. (2000) PNAS 97:925-930.
A method to identify NSAID derivatives and NSAID analogues that possess A(342 lowering ability and have altered COX activity is described. NSAID derivatives and NSAID analogues of aminocarboxylic acids, arylacetic acids and arylpoprionic acids can be tested for their ability to lower A(342 and increase A(338 in cultured cells and in animals (as described herein). They also can be tested simultaneously for their ability to inactivate COX-1 and COX-2 using ih vitro assays as described by Kalgutkar et al.
(2000) PNAS 97:925-930. Derivatives of the NSAIDs sulindac, meclofenamic acid, flufenamic acid, indomethacin, carprofen, fenoprofen, and flurbiprofen that can be tested include the following:
(1) meclofenamic acid and flufenamic acid derivatives in which (a) the position of the carboxylic acid substituent on the phenyl ring is altered, (b) the position or type of substituents on the phenyl ring opposite the caraboxylic acid substituent are altered, (c) the bond connecting the two phenyl rings is altered, (d) the carboxylic acid substituent is altered to a propionic acid or other derivative, or (e) any combination of these alterations;
(2) fenoprofen, flurbiprofen, and carprofen derivatives in which (a) the position of the propionic acid substituent on the phenyl ring is altered, (b) the position or type of substituents on the phenyl ring opposite the propionic acid substituent is altered, (c) the bond connecting the two phenyl rings is altered, (d) the acetic acid substituent is altered to a carboxylic acid or other derivative, or (e) any combination of these alterations;
(3) indomethacin derivatives in which (a) the carboxylic acid group on indomethacin is altered to other substituents, (b) the substituent on the indole nitrogen is altered, or (c) any combination of the two;
(4) sulindac sulfide in which (a) the methylthio derivative of sulindac sulfide is altered to other substituents, (b) the propionic acid derivative is altered to other substituents, (c) the Fluoride is altered to other substituents, or (d) any combination of the above.
In addition structural analogues of NSAIDs that possess A(342 lowering ability, identified by pharamacophore searches (Perola et al., (2000) J. Med Chem.43:
401-408) or other computer based structural comparison programs of commercially available compounds can be tested for A[34a lowering activity, ability to increase A(338, and COX
inhibition as described herein.
3. Identification of mammals in heed of treatment with an A,li4~ lowering agent Clinical symptoms of AD include, for example, progressive disorientation, memory loss, and aphasia; eventually, disablement, muteness, and immobility occur.
Pathological indicators of AD include, for example, the presence of neurofibrillary tangles, neuritic plaques, and amyloid angiopathy. Preventing the progression of AD can be interpreted to mean preventing the onset or further development of clinical symptoms and/or pathological indicators of AD. For example, an individual who does not have clinical symptoms or pathological indicators of AD can be prevented from developing clinical symptoms or pathological indicators. Further, an individual who has a mild form of AD can be prevented from developing a more severe form of AD. Delaying the progression of AD can be interpreted to mean delaying the time of onset of AD-related symptoms andJor pathological indicators or slowing the rate of progression of AD, determined by the rate of development of clinical symptoms and pathological indicators.
Reversing the progression of AD can be interpreted to mean that the severity of an AD
condition has been lessened, i.e., the AD condition of an individual has changed from severe to less severe as indicated by fewer clinical symptoms or pathological indicators.
An individual can choose to take an A(34~ lowering agent as a preventative measure to avoid developing AD. For example, an individual with a genetic predisposition to AD can take an A[342 lowering agent to prevent or delay the development of AD. A genetic predisposition can be determined based on known methods. For example, an individual can be considered to have a genetic predisposition to AD if the individual has a family history of AD. Genetic predisposition to AD also can include point mutations in certain genes such as the APP gene, the presenilin-1 or presenilin-2 gene, or the apolipoprotein E gene. Such mutations can predispose individuals to early-onset familial AD (FAD), increased risk of developing AD, or decreased age at onset of AD. (See page 1332, Table 30-2 of Cotran et al.
(1999) Robbins Pathologic Basis of Disease, Sixth Edition, W.B. Saunders Company; and U.S.
Patent No. 5,455,169.) Furthermore, an individual who has clinical symptoms of, or has been diagnosed with, AD can take an A[342 lowering agent to prevent or delay fuxther progression of AD as well as to reverse the pathological condition of the disease.
An AD diagnosis can be made using any known method. Typically, AD is diagnosed using a combination of clinical and pathological assessments. For example, progression or severity of AD can be determined using Mini Mental State Examination (lVllVISE) as described by Mohs et al. (1996) Int Psychoge~iatr 8:195-203;
Alzheimer's Disease Assessment Scale- cognitive component (ADAS-cog) as described by Galasko et al. (1997) Alzheimer Dis Assoc DisoYd, 11 suppl 2:533-9; the Alzheimer's Disease Cooperative Study Activities of Daily Living scale (ADCS-ADL) as described by McKhann et al. (1984) Neurology 34:939-944; and the NINCDS-ADRDA criteria as described by Folstein et al. (1975) JPsychiatr Res 12:189-198. In addition, methods that allow for evaluating different regions of the brain and estimating plaque and tangle frequencies can be used. These methods are described by Braak et al. (1991) Acta Neuropathol 82:239-259; Khachaturian (1985) Arch Neuro 42:1097-1105; Mirra et al.
(1991) Neurology 41:479-486; and Mirra et al. (1993) Arch Pathol Lab Med 117:132-144.
4. Treatment of mammals with A~342 lowering agents A[342 lowering agents can be administered in any standard form using any standard method. For example, A(342 lowering agents can be in the form of tablets or capsules that are taken orally. A[342 lowering agents also can be in a liquid form that can be taken orally or by injection. A~i4a lowering agents also can be in the form of suppositories.
Further, A(34a lowering agents can be in the form of creams, gels, and foams that can be applied to the skin, or in the form of an inhalant.
A(34a lowering agents can be administered at any dose that is sufficient to reduce levels of A[342 in the blood plasma, CSF, or brain. Lower doses can be taken over a period of years to prevent and/or delay the progression of AD. Higher doses can be taken to reverse the progression of AD. Depending on the effectiveness and toxicity of a particular A(34a lowering agent, an A(34a lowering agent can be used at a dose of 0.1-50 mg/kg/day.

5. Compositions a~zd kits The invention also provides pharmaceutical compositions containing combinations of an A(342 lowering agent and an antioxidant effective in preventing, delaying, or reversing the progression of Alzheimer's disease. An A(342 lowering agent of the invention that has the ability to reduce A(342 levels can be combined with any antioxidant. The antioxidant can be a vitamin, for example vitamin E, vitamin C or curcumin; the antioxidant also can be Gingko biloba. Other pharmaceutical compositions can include an A~i42 lowering agent and a non-selective secretase inhibitor or an acetylcholinesterase inhibitor.
The pharmaceutical composition can be in any form, for example tablets, capsules, liquids, creams, gels, or suppositories and can include a suitable pharmaceutical carrier. In addition, the invention provides kits containing pharmaceutical compositions of A(34a lowering agents and antioxidants as well as instructions that indicate dose regimens for effective use.
The invention will be further described in the following examples, which do not limit the scope of the invention described in the claims.
EXAMPLES
Example 1 - Cell cultures, drug treatments, and cell toxicity analysis Cell cultures were maintained in standard cell culture media supplemented with 10% fetal bovine serum and 100 UImL penicillin/streptomycin (Life Technologies Inc., Germany). Cell cultures consisted of the following: Chinese hamster ovary (CHO) cells that expressed human APP751 from a vector containing a gene encoding APP75I;
CHO
cells that expressed both human APP751 and human mutant PS-1 (M146L) from vectors containing genes encoding APP751 and mutant PS-1 (M146L); CHO cells that expressed human mutant APP751 (V717F) from a vector containing a gene encoding mutant APP751 (V717F); human neuroglioma cells HS683 that expressed human APP695 from a vector containing a gene encoding APP695; HEK 293 cells that expressed human APP695 from a vector containing a gene encoding APP695; and embryonic fibroblasts (that had immortalized spontaneously) from COX-1 and COX-2 double-knockout mice.
The NSAIDs, sulindac sulfide (50 mM, Biomol, PA, USA), sulindac sulfone (50 mM, Biomol, PA, USA), naproxen (100 mM, Cayman Chemical, MI, USA), and aspirin (2.5 M, ICN Biomedicals, CA, USA) were dissolved in the vehicle DMSO.
Indomethacin (50 mM, Biomol, PA, USA) and (S)-ibuprofen (250 mM, Biomol, PA, USA) were dissolved in ethanol. Celecoxib and rofecoxib capsules were obtained from and dissolved in ethyl acetate. For ar_alyses of A(3 secretion, APP
processing, and notch cleavage, cells were cultured in serum-containing media and pretreated overnight with a specific NSAID. The next day, media were changed and cultures were treated with the same NSAID for another 24 hours.
NSAID toxicity in CHO or HS683 cells was examined using standard MTT-assay (3-(4,5-Dimethyl-2-thiazolylyl)-2,5-diphenyl-2H-tetrazolium Bromide) or [3H]-thymidine incorporation assay. For cell toxicity studies, cells were treated with sulindac sulfide at concentrations up to 100 p,M, indomethacin at concentrations up to 200 ~,M, and ibuprofen at concentrations up to lmM.
Example 2 - Antibodies Antibodies used included the following: SA3 and 1G7, two monoclonal antibodies that recognized non-overlapping epitopes between residues 380-665 of APP;
CT15, a polyclonal antibody that recognized the C-terminal fifteen amino acid residues of APP;
26D6, a monoclonal antibody that recognized amino acid residues 1-12 of the A(3 sequence; 9E10, a monoclonal antibody that recognized the myc-epitope sequence; anti-COX-2 antibody, a monoclonal antibody that recognized COX-2; and M-20, a polyclonal antibody that recognized COX-1. The antibodies SA3, 1G7, CT15, and 26D6 were described by Koo et al. (1996) J Cell Sci 109:991-8; Sisodia et al. (1993) JNeurosci 13:3136-42; and Lu et al. (2000) Nat Med 6:397-404. The monoclonal antibody was purchased from Calbiochem-Novobiochem, CA, USA. The monoclonal anti-COX-2 antibody was purchased from BD Transduction Laboratories, CA, USA. The polyclonal antibody M-20 was purchased from Santa Cruz Biotechnology, CA, USA.
Example 3 - ELISA
A(3 was detected by sandwich enzyme-linked immunosorbent assay (ELISA) as described by Murphy et al. (2000) JBiol Chem 275:26277-84. Following NSAID
treatment, culture supernatants were collected, and cell debris was removed by centrifugation. Complete protease inhibitor cocktail (Roche Molecular Biochemicals, IN, USA) was added to the media and A(34o and A~342 levels were quantified using end-specific A(3 ELISAs. All measurements were performed in duplicate.
Example 4 - Adenoviral infection of emb~-yohic f broblasts derived from COX

double-knockout mice The adenoviral vector containing a gene encoding APP695 was described by Yuan et al. (1999) JNeurosci Methods 88:45-54. Primary fibroblasts derived from COX-1/COX-2 double-knockout mice were infected with 100 plaque-forming units (PFU) of viral vector per cell. Infection was performed in serum-free medium for two hours.
Medium was changed and cells were treated with NSAIDs as described in Example 1.
Example 5 - Analyses ofAPP and Notch processing Expression of holo-APP and APP C-terminal fragments (CTFs) was examined by Western blot analysis using antibody CT-15. APP secretion was examined by Western blotting using a mixture of SA3/IG7 antibodies. APP turnover was examined by pulse labeling of CHO cells with 150 ~,Ci [35S]-methionine for fifteen minutes followed by a cold chase step for up to four hours. Cell lysates were immunoprecipitated with antibody CT-15, subjected to SDS-PAGE, and analyzed by phosphor imaging.
APP surface expression and internalization were measured as described by Koo et al. (1996) JCell Sci 109:991-8. Iodinated antibody 1G7, at approximately 3-6 ~.Ci/~g, was applied to confluent layers of CHO cells in binding medium (DMEM, 0.2%
BSA, 20 mM HEPES [pH 7. 4]) and incubated at 37 °C for thirty minutes. After incubation, cells were rapidly chilled on ice and the reaction was quenched by the addition of ice-cold binding medium. To remove unbound antibody, chilled cells were washed multiple times with ice-cold Dulbecco's phosphate-buffered saline (Life Technologies Inc.).
Antibody bound to cell surface APP was detached by washing with ice-cold PBS (pH 2) for five minutes; this constituted the acid-labile APP antibody pool. Cells were lysed in 0.2 M
NaOH; lysates contained the acid-resistant APP antibody pool. Acid-labile and acid-resistant APP antibody counts were measured by y counting. The ratio of acid-resistant to acid-labile count was a measure of the internalized to the cell surface APP
pool.
Two Notch-encoding vector constructs were used in examining Notch processing.
These were a construct expressing a myc-tagged NHa-terminal truncated Notch-1 polypeptide (Notch~EMV), and a construct expressing only the Notch intracellular cytoplasmic domain (hTICD) (see Kopan et al. (1996) Proc Natl Acad Sci USA
93:1683-8). In the construct expressing a myc-tagged NHa-terminal truncated Notch-1 polypeptide, the start codon, a methionine at position 1726, was mutated to a valine to eliminate translation initiation.
Example 6 - Mass spectrometry Secretion of A(3 peptides was analyzed using immunoprecipitation/mass spectrometry as described by Wang et al. (1996) JBiol Chem 271:31894-902.
Briefly, 1 mL amount of culture supernatant was subjected to immunoprecipitation using the monoclonal antibody 4G8 (Senetek, CA, USA). Molecular masses and concentrations of A[3 peptides were measured using a matrix assisted laser desorption/ionization time-of flight (MALDI-TOF) mass spectrometer. To compare the concentrations of individual A(3 species in culture supernatants, synthetic A(312_a8 peptides (Sigma, MO, USA) were added to the supernatant samples as internal standards and relative peak heights were calculated.
Example 7 - BicinelUrea A,13 western blot analysis Bicine/LTrea A(3 western blot analysis was performed as described by Wiltfang et al. (1997) Electrophoresis 18:527-32. A 1 mL amount of culture supernatant was subjected to immunoprecipitation using monoclonal antibody 26D6.
hnmunoprecipitates were mixed with sample buffer and heated to 95 °C for five minutes.
Eluant samples were separated on BicinelLTrea gels, then transferred to nitrocellulose membranes, and probed with antibody 26D6. Standard A[31_40, A(3i_42 and A[31_3s peptides (Sigma, MO, USA) were used for identification of the A~3 species.
Example 8 - Cells treated with the non-selective COX inhibitor sulindac sulf de showed reductions in levels of A,~342 secretions Cell cultures were treated with increasing concentrations of the NSAID
sulindac sulfide. Levels of A(34o and A(342 in culture supernatants were analyzed using ELISA.
Figure 1 is a graph comparing the A(342/A(34o ratios of sulindac sulfide-treated CHO cell cultures expressing APP751 and the PS-1 mutant M146L. A(34a/A(34o ratios and total A(3 levels (i.e., the sum of A[34o and A(342 values) were normalized to values obtained from DMSO-treated cells. Results shown were averages of two or three experiments performed in duplicate. CHO cell cultures treated with 40-60 ~,M sulindac sulfide showed a 50% reduction in A(34z/A[34o ratios. No significant reduction in total A(3 level was observed. Therefore, treatment of CHO cells expressing APP and mutant PS-1 with the NSAID sulindac sulfide reduced the A(342/A(34o ratio by selectively reducing A(34a secretion in a dose-dependent manner. This was confirmed in CHO cells that expressed wild type APP751 as well as those that expressed mutant APP V717F (data not shown).
To rule out potential cell type-specific effects, A(3 secretion in response to sulindac sulfide treatment was examined in the human neuroglioma cell line HS683 that expressed APP695. Figure 2 is a graph comparing A(342/A(34o ratios in HS683 cells expressing APP695 that were treated with DMSO with those of cells treated with various concentrations of sulindac sulfide. A dose-dependent reduction of A(342 secretion, similar to that exhibited by CHO cells, was observed. Sulindac sulfide also reduced A(34a secretion in kidney HEK293 cells and primary mouse embryonic fibroblasts (data not shown). No cell toxicity was observed at sulindac sulfide concentrations up to 100 p.M
(data not shown).
Example 9 - Cells heated with other holyselective COX inhibitors such as ibuprofen and indomethacin showed r0ductions in levels ofA,(342 secretion Cell cultures were treated with increasing concentrations of the NSAIDs ibuprofen and indomethacin. A(34o.and A(342 levels in culture supernatants were analyzed using ELISA as described in Example 3. Figures 3 and 4 are graphs comparing A[342/A(34o ratios observed for CHO cells expressing APP751 and the PS-1 mutant M146L when treated with various concentrations of ibuprofen and indomethacin, respectively.
A[34a/A[34o ratios and total A(3 levels were normalized to values obtained from ethanol-treated cells. Results shown were averages of two or three experiments, each performed in duplicate. Dose dependent reductions in A~i42/A(34o ratios by selective reductions of A(34a secretion were observed for both ibuprofen and indomethacin. A 50%
reduction in the A(342/A(34o ratio was reached at ibuprofen concentrations between 200-300 ~,M and at indomethacin concentrations between 25-50 ~.M. Total A(3 levels were not significantly affected at ibuprofen concentrations up to 500 ~,M (see Figure 3) and at indomethacin concentrations up to 100 ~.M (see Figure 4). No cell toxicity was observed in CHO cells treated with ibuprofen concentrations up to 1mM or indomethacin concentrations up to 200 p,M (data not shown).
Example 10 - Reduction of A/342 secretion is hot associated with COX
inhibitory activi or with all NSAIDs The effect of sulindac sulfone on A(342 secretion was examined. Sulindac sulfone is an oxidation product of the pro-drug sulindac. Like sulindac sulf de, sulindac sulfone inhibits proliferation and induces apoptosis in human cancer cell lines in vitro (see Piazza et al. (1995) CancerRes 55:3110-6). In contrast.to sulindac sulfide, sulindac sulfone is devoid of any inhibitory effect on COX. Cell cultures were treated with increasing concentrations of sulindac sulfone. A[34o and A(342 levels in culture supernatants were analyzed using ELISA. When CHO cells expressing APP 751 were treated with sulindac sulfone, no changes in A[342/A(34o ratios were observed with sulindac sulfone concentrations of up to 400 ~,M (data not shown). The inability to reduce A[34a secretion by the non-COX-inhibitor sulindac sulfone suggested an important mechanistic role for COX inhibition in the selective inhibition of A(342 secretion by NSAIDs.
To determine whether reduction of A~i4a secretion is a common effect of all NSAIDs, other clinically useful NSAIDs were examined. Naproxen is a non-selective COX-inhibitor with an inhibition profile similar to sulindac and a structure similar to ibuprofen (see Cryer et al. (1998) Am JMed 104:413-21). Cell cultures were treated with increasing concentrations of naproxen and aspirin. A(34o and A(342 levels in cell culture supernatants were analyzed using ELISA. A(342/A(34o ratios and total A(3 levels were normalized to values obtained from DMSO-treated cultures. Averages of two or three experiments performed in duplicate are summarized in Figure 5. Treatment of CHO cells expressing APP751 with naproxen, at concentrations up to 400 p,M, did not change A(342/A(34o ratios and did not affect total A(3 levels (see Figure 5).
Similarly, no reductions in A(342 secretion were observed when cultures were treated with aspirin concentrations of up to 3 mM (data not shown). Two selective inhibitors of COX-2, celecoxib and rofecoxib, also were examined to determine if they reduced A/342 secretion.
Celecoxib and rofecoxib were prepared from capsules using solvent extraction and recrystallization. NSA)Ds were verified using NMR and mass spectrometry. CHO
cells expressing APP751 were treated with various concentrations of celecoxib.
Figure 6 is a bar graph comparing A(342/A[34o ratios and total A(3 levels in cells treated with ethyl acetate or various concentrations of celecoxib. Results showed that 20~,M
celecoxib treatment induced a two-fold increase in A(342/A(34o ratio. The increase in A(342/A[34o ratio also was observed when human neuroglioma cells were tested (data not shown).
The increase in A(342/A(34o ratio was not seen in cells treated with rofecoxib at 20 pM (data not shown). Diclofenac and NS-398, two other NSAIDs having preferential activities against COX-2, did not affect A(342/A(34o ratios or total A(3 levels. Table 1 summarizes selective and non-selective COX-inhibitors that were tested and results of these tests.
Reduction of A(34a secretion was not associated with all NSAIDs. (Note: peak NSA)D
concentrations used in these experiments were higher than what was required for complete inhibition of COX-1 and COX-2 activities in in vitro cell-based assays.) Table 1: Non-selective and selective COX-inhibitors tested for effect on A(342 levels Drug Highest conc. testedPlasma A[342/Aj340 ratio COX-1/COX-2 selectivity#

Non-selective COX-inhibitors (1=equal activity) Sulindac sulfide 100 14.6 (gM) selective decrease0.61 (~,M) in A~i42 Indomethacin 150 1.4 selective decrease22-58 in A[342 Ibuprofen 750 40-111 selective decrease1.69 in A[342 Naproxen 400 1.3 no effect 1.79 Aspirin 3000 111 no effect 166 Meloxicam 100 15 no effect .O1-0.3 Diclofenac* 600 6.1 no effect .69 Selective COX-2 inhibitors NS-398* 20 no effect .07 Celecoxib 20 (gM) 15 (nM) selective increase.003 in A(342 Rofecoxib* 20 (~M) 3 (nM) no effect .001 To confirm that NSA)D did not reduce A(34z secretion through COX inhibition and though reduction of prostaglandin synthesis, cells devoid of COX-1 and COX-2 activities were treated with sulindac sulfide, and A[342/A(34o ratios were examined.
Primary fibroblasts derived from COX-1/COX-2 double-knockout mice, described by Zhang et al.

(1999) JExp Med 190:451-59, were infected with an adenovirus vector that encoded APP695 (see Yuan et al. (1999) JNeu~osci Methods 88: 45-54). Fibroblasts infected with the adenovirus vector expressing APP695 were treated with increasing concentrations of sulindac sulfide. Levels of A(3 forms in fibroblast culture supernatants were quantified using ELISA and results are summarized in Figure 7.
(A(34z/A[34o ratios and total A[3 levels were normalized to values obtained from DMSO-treated cells.
Results were the averages of two or three experiments, each performed in duplicate.) Sulindac sulfide reduced A[34z secretion as well as the A(34z/A(34o ratio of fibroblasts in a fashion similar to that seen with CHO and HS683 neuroglioma cells. Therefore, selective reduction of A~34z was not mediated by COX inhibition.
Example 11 - APP processing by a and ,~i-secretases, APP turnover, and notch intramembrane cleavage are not a~'ected by sulindac sulfide NSAIDs are the only compounds reported so far that change A(34z/A/34o ratios by selectively decreasing A~iaz secretion. To determine if APP processing and notch intramembrane cleavage were affected in cells treated with NSAIDs, the following experiments were performed.
CHO cell cultures expressing APP751 were treated with increasing concentrations of sulindac sulfide. CeII lysates were prepared, and steady-state APP levels were examined using 4-12 % gradient-gel electrophoresis and western blotting using the polyclonal antibody CT15. When western blot analysis was performed, neither a change in APP levels, nor an increase in CTF levels was observed in response to 60 p,M or 80 pM sulindac sulfide treatment compared with levels observed for cells treated with DMSO. Unlike published y-secretase inhibitors, sulindac sulfide did not induce detectable accumulation of APP CTFs. Therefore, [3-secretase cleavage was not significantly affected by sulindac sulfide.
When western blot analysis was performed to detect soluble APP (sAPP) in culture supernatants using SA3/IG7 monoclonal antibodies, results showed that there was no significant change in secretion of the APP ectodomain, (i.e., sAPP), in response to increasing concentrations of sulindac sulfide. Therefore, a-secretase cleavage was not significantly affected by sulindac sulfide.

APP turnover in the presence of sulindac sulfide was examined by (1) pulse labeling CHO cells with 35S-methionine and (2) determination of APP half life.
All values were normalized to a signal obtained at the end of pulse labeling. When the APP
half life in cells treated with DMSO was compared with APP half life in cells treated with sulindac sulfide at 25 or 125 p,M, APP half life after treatment with 25 or 125 ~,M
sulindac sulfide was similar to APP half life after treatment with DMSO.
Therefore, APP
turnover was not altered significantly in the presence of sulindac sulfide.
A significant fraction of A[3 is produced and released in the endocytic pathway after internalization of APP from the cell surface (see Koo et al. (1994) JBiol Chem 269:17386-9). The effect of sulindac sulfide on this endocytic pathway was examined with an APP internalization assay described by Koo et al. (1996) J Cell Sci 109:991-8.
APP internalization was expressed as a ratio of cell surface APP versus internalized APP.
When APP internalization in cultures treated with DMSO was compared with APP
internalization in cultures treated with sulindac sulfide at 60 or 80 p,M, the ratio of cell surface APP to internalized APP was not altered in cells treated with sulindac sulfide compared to cells treated with DMSO alone. Therefore, it was concluded that APP
internalization was unchanged after sulindac sulfide treatment.
Notch intramembrane cleavage and formation of NICD were analyzed in kidney HEK293 cells. The myc-tagged Notch~lEMV construct encoding a constitutively cleaved Notch variant was transiently transfected into HEK293 cells. Cell cultures were treated with 125 ~,M sulindac sulfide for 36 hours. Then they were pulse labeled with methionine for thirty minutes and chased for two hours. Cell lysates were prepared and subjected to immunoprecipitation with monoclonal antibody 9E10.
Immunoprecipitated proteins were subjected to SDS-PAGE and phosphor imaging analyses. When amounts of NICD immunoprecipitated from lysates of cells treated with DMSO were compared with amounts immunoprecipitated from lysates of cells treated with sulindac sulfide, results showed that treatment with sulindac sulfide did not impair Notch cleavage and NICD formation. (Cells transfected with a construct encoding only the NICD
domain were used for identification of the cleavage fragment.) Similarly, treatment with 500 p,M
ibuprofen or 150 p,M indomethacin did not cause accumulation of APP-CTFs or inhibition of Notch cleavage (data not shown). Overall, these results demonstrated that NSAID treatment did not significantly perturb APP processing or y-secretase activity.
This, however, did not rule out modulation of y-secretase activity as a mechanism of action for NSAIDs. The selective reduction in A(34a secretion could be reflected only in minor changes of y-secretase activity that may not be detectable in the assays described above.
Example 12 - Reduction in A,1342 secretion was accompanied by a dose-dependent increase in Aal_38 SpecEes To examine A(3 species secreted by cells treated with sulindac sulfide, immunoprecipitation and mass spectrometry analyses were performed. Figure 8 is two representative mass spectra of A[3 species secreted by CHO cells expressing APP751 after treatment with DMSO or after treatment with 100 ~.M sulindac sulfide. After treatment with 75-100 ~,M sulindac sulfide, a strong reduction in A(34a secretion was observed.
Levels of A(34o, however, were largely unaffected. Various A(3 species including A(31~.a, 1 S A(31_39, Ay-3s, and Ay-s~ were quantified. Figure 9 is a bar graph comparing ratios of each of these species to A~il~o, i.e. A(31_X/A(31-4o ratios, at 75 or 100 ~,M
sulindac sulfide.
Duplicate measurements were used in generating the bar graph. Reductions in A[34~/A[34o ratios were accompanied by two-fold increases in A(31_38/A(31~o ratios.
Increases in A(31_3$
levels were dose-dependent. Other A(3 peptide levels did not vary consistently between cells treated with DMSO or with sulindac sulfide.
Mass spectrometry results demonstrating reductions in A/342 secretion with concomitant increases in A(31_38 secretion were confirmed by immunoprecipitation. A(3 polypeptides were immunoprecipitated from culture supernatants of CHO cells expressing APP751 and mutant PS-1. Immunoprecipitates were separated on an SDS-urea gel system that can resolve individual A[3 species (see Wiltfang et al.
(1997) Electrophoresis 18:527-32). Standard A(31_38, A/31_40, and A(31_4a peptides were included for identification of different A(3 species. When changes in A(338, A(34o, and A(3az levels in CHO cells treated with DMSO were compared with those in cells treated with 60 p,M
or 80 ~,M of sulindac sulfide, a reduciion in the intensity of an immuno-reactive band corresponding to A(342 was observed. This reduction was matched by an equivalent increase in the intensity of an immuno-reactive band corresponding to A(31_3s.

Two potential mechanisms may explain this unprecedented change in A(3 production after NSAID treatment. Sulindac sulfide could reduce A(342 secretion by shifting y-secretase activity towards production of A(31_38. Alternatively, it may stimulate a novel proteolytic activity that converts A(34a into shorter A(3 species such as A(31_3s.
Koo et al. (1994) JBiol Chem 269:17386-9 and others reported that APP
processing in the endocytic pathway leads to the generation and release of both A[34o and A(34a into culture supernatant. To examine the intracellular pool of A(34a in APP mutants that lack the endocytic signal, CHO cells expressing an internalization-deficient APP
polypeptide lacking 43 amino acids in the cytoplasmic tail were used (Perez et al. (1999) JBiol Chem 274:18851-6). Levels of cellular and secreted A(34a and A(34o in cells expressing wild type APP and in cells expressing mutant APP were compared using ELISA. Results indicated that in the absence of the cytoplasmic tail, levels of A[34o and A~342 secreted by cells expressing mutant APP were diminished compared to cells expressing wild type APP. In addition, in the absence of a cytoplasmic tail, cellular A(34o levels were reduced while cellular A(342 levels were not reduced.
Example 13 - NSAID treatment of Tg2576 tYa~asgefaic mice NSAIDs were dissolved in an appropriate vehicle. Dimethyl sulfoxide (DMSO), ethanol, and ethyl acetate are some examples. The NSAID solution was mixed with Kool-Aid and administered orally using a medicine dropper. For three days, equal doses were administered every four hours, totaling 50 mg/kg/day. At two hours after the final doses were administered, animals were sacrificed, and SDS soluble A[34o and A(342 were analyzed using ELISA.
Example 14 - Treatment of animals with ibuprofen reduces A~i42 levels To determine whether acute ibuprofen treatment of mice would reduce A(34a levels, three month-old Tg2576 mice expressing APP695 containing the 'Swedish"
mutation (APP695NL) were used. Three month old mice have high levels of soluble A[3 in the brain but no A[3 deposition (see Hsiao et al. (1996) Science 274:99-102). Mice were given naproxen, ibuprofen, or meclofenamic acid as described in Example 13. Mice treated with ibuprofen (n=12) were compared with those untreated (n=11), treated with naproxen (n=7), or treated with meclofenamic acid (n=4). Brain levels of SDS-soluble A(34o and A(342 were measured using ELISA. Table 2 summarizes A(34o and A[34a levels determined for the control group and the naproxen, ibuprofen, and meclofenamic acid-treated groups. Treatment with ibuprofen or meclofenamic acid for three days resulted in S approximately 30% reduction in A[342 levels in the brain, while no change was observed in A~i4o levels (see Figure 10). No reduction in A[342 levels was observed for naproxen-treated mice. These data were consistent with the rapid onset of A(342 reduction in cell culture studies and illustrated that cell culture experiments were able to predict in vivo efficacy. In addition, these data suggested that ibuprofen treatment could prevent amyloid pathology by decreasing A(342/A[34o ratio in the brain.
Table 2: Brain levels of A(3 after acute dosing of Tg2576 mice (mean ~ SD) Meclofenamic Control (n=11)Naproxen Ibuprofen (n acid (n =
(n = 7) = 12) 4) A(340 (finol/gm)2603 314 2786 179 2620 246 2932 289 A~342 1074 t 145 1182 93 734 302 * 679 343 **

%A(342 I 29.32.9 29.81.6 21.57.7* 18.68.7**

* =p < 0.05; ** =p < U.U1, Dunnett's test 1 S Example 1 S - NSAIDs, NSAID derivatives, and NSAID analogues NSAIDs that are screened for the ability to reduce A(34z levels include: FDA-approved NSAIDs, NSAIDs derivatives, and NSAID analogues most potent for reducing A(342 levels, newly synthesized derivatives and analogues of the most potent NSAIDs, and NSAIDS known to taxget pathways other than COX pathways. FDA-approved NSAIDs include ibuprofen, naproxen, dicolfenac, aspirin, indomethacin, fenoprofen, flurbiprofen, ketorolac. Derivatives of the most potent NSAIDs include aryl propionic acid derivatives such as ibuprofen and fenoprofen, and the anthranilic acid derivatives (also called amino carboxylic acid derivatives) such as the meclofenamic acid series and flufenamic acid.
(NSAIDs in both series share a similar core structure of either a diphenyl ketone or 2S dephenyl ether.) Other derivatives or analogues that are screened for the ability to reduce A~i4a levels include flufenamic acid, indomethacin, and meclofenamic acid derivatives and analogues (see Figure 11 and Kalgutkar et al. (2000) J of Med Chem 43:2560-70).
Newly synthesized NSAID derivatives or analogues include novel biphenyl amines (Figure 12) and diphenyl ketones. Examples of NSAms that target additional pathways to COX include LOX inhibitors.
Once a set of NSAIDs, NSAID derivatives, or NSAID analogues having potent ability to reduce A~i4a levels is obtained, a pharmacophore search is performed to identify other NSAIDs structurally similar to those in the set. If a large number of candidates are identified, the structurally similar NSAIDs are subjected to a secondary structural screen using a computer-based molecular docking algorithm known as EUDOC. In the second structural screen, crystal structures and COX-1l COX-2 binding pockets are used to identify a subset consisting of NSAIDs structurally similar to those that have potent ability to reduce A(342 levels but do not bind COX-1 or COX-2. NSAIDs predicted to bind to COX and those predicted to not bind to COX are used as controls.
NSAIDs, MAID derivatives, and NSAID analogues can be obtained commercially or they can be chemically synthesized. Novel NSAIDs, NSAID
derivatives, or NSAID analogues with unknown effects on COX activity are tested using in vitro COX-l and COX-2 assays to determine if there is an affect on COX
activity.
Commercially available kits from Oxford biochemicals are used for COX
inhibition assays.
Example 16 - Determination of optimal screening interval foY detecting selective seduction of X342 levels To determine the optimal treatment interval for examining selective reduction of A(34a levels, CHO-APP695NL,I,his cell cultures were treated with the vehicle, or treated with ibuprofen or meclofenamic acid for six, twelve, or twenty-four hours.
A~i4o and A(342 levels in culture supernatants were determined for each time points using ELISA.
Figure 13 is a bar graph demonstrating that selective reduction of A~i4a was detectable at six hours when cells were treated with meclofenamic acid. Similar results were observed for ibuprofen (data not shown).

Example 17 - Primary in vitro screening In a primary screen, the effects of NSAIDs on A(342 secretion by a CHO cell line that expressed APP (CHO-APP69SNL,I,his) were examined. Duplicate cell cultures were treated with (a) a vehicle, (b) 10 ~.M of NSAID, or (c) 100 ~.M of NSAID.
S To determine A(34o and A(342 levels, six-hour culture supernatants taken from cells grown in a single well of a twenty-four-well plate were used in end-specific A(34o and A(34a ELISAs (Suzuki, et al. (1994) Sci 264:336-1340). A(34o and A(342 levels of cultures treated with NSAID were compared with those of cultures treated with the vehicle alone.
Concentrations of 100 ~.M Ibuprofen and 10 ~.M meclofenamic acid were used as positive controls. Results, in Table 3, indicated that some NSAIDs selectively reduce A(34a levels, but at the concentrations tested, many do not. NSAIDs were classified based on a 20%
change in A(3 levels observed in NSAID-treated versus vehicle treated cells.
Classification was made based on a 20% change because the data showed a 10%
accuracy variance. When classification was made based on a 20 % change, all NSAIDs screened, 1 S with the exception of two, were classified in the same category with repeated testing.
Two NSAIDs, shown in bold italic, gave results that altered their categorization upon re-screening; classification was resolved after a third test. These results confirmed the data described in Examples 8-10, as the NSAIDs that were shown to selectively lower A(34a initially also reduced A[342 in this screen. Of the newly synthesized biphenyl amines, meclofenamic, mefenamic, and flufenamic acid selectively reduced A(342 levels, while tolfenamic acid did not. NSAIDs that caused either selective reduction of A(342 levels or reduction in both A(34o and A(342 levels are subjected to a secondary screen.

Table 3. Effects of NSAIDs on secreted A(3.
Type %Co~:tro%Cohtro%Cohtro Compound A/~40A/742%A,Q42 .LA 42 no effect on A 40 Sulindac SulfideCox-1,2 97% 57% 65%
p,M

Flufenamic AcidCox-1,2 99% 64% 70%
10 pM

Ibuprofen 100 Cox-1,2 95% 74% 81%
wM

Ibuprofen 10 Cox-1,2 102% 80% 82%
~.M

Flurbiprofen Cox-1,2 93% 70% 80%
100 pM

Fenoprofen 100 Cox-1,2 102% 60% 63%
p,M

Mefenamic Acid Cox-1,2 116% 78% 72%
100 wM

Indomethacin Cox-1,2 101% 69% 68%
100 pM

~A 42 > ~A 40 NPPB 10 Etm Cox-1,2 81% 48% 66%

Carprofen 100 Cox-1,2 58% 48% 86%
pM

Meclofenamic Cox-1,2 39% 13% 37%
Acid 10 pM

.~A 40 no effect on A 42 APHS 101tM Cox2>Coxl50% 114% 178%

Resveratrol Cox -1 75% 107% 130%
10 pm .~A 40 and TA

Meloxicam 10 Cox-1,2 64% 122% 158%
pM

SC560 10 pM Cox-1>Cox47% 166% 227%

Guaiazulene Cox-1,2 70% 124% 156%
100 pM

NS398; 10 pm Cox- 2 101% 146% 132%
> Cox-1 Ketorlolac IO Cox-1,2 g4% 131% 142%
pM

Benzydamine Cox-1,2 90% 128% 132%
100 ~,M

TA 40 and/or Suprofen 100 Cox-1,2 126% 129% 102%
pM

Indoprofen 100 Cox-1,2 116% 126% 107%
p.M

Nabumetone 100 Cox-1,2 157% 103% 70%
pM

Piroxicam 100 Cox-1,2 142% 101% 75%
p,M

o Effect on A

AcetylsalicylicCox-1> 93% 99% 104%
acid 100 p,M Cox-2 Ketoprofen 100,ttMCox-1,2 88% 107% 117%

Fenbufen 100 Cox-1,2 100% 109% 107%
p,M

Naproxen 100 Cox-1,2 107% 112% 104%
~,M

Isoixicam 100 Cox-1,2 109% 112% 103%
uM

Tenoxicam 100 Cox-I,2 g0% 92% 112%
p,M

Tolfenamic AcidCox-1,2 g4% 95% 110%
100 ~,M

Diclofenac; Cox-1,2 gg% 87% 100%
100 p.m Etodolac 100 Cox-1,2 g5% 109% 120%
wM

Acemetacin 100 Cox-1,2 110% 101% 93%
pM

Niflumic Acid Cox-1,2 120% 107% 85%

Dapsone Anti -Bacterial99% 80% 84%

Sulindac SulfoneNo-Cox 109% 97% 92%

Nimesulide Cox-1,2 105% 116% 116%

Suxibuzone Cox-1,2 82% 107% 129%

Diflunisal Cox-I,2 90% 103% 112%

Example 18 -Secondary and tertian in vitro NSAID screening In a secondary screen, an extended dose-response study in which CHO cell cultures are treated with 1nM to 1mM of NSAID is performed. Dose response studies are used to estimate ICso values for maximum reduction of A(3 levels as well as to identify NSAIDs that have toxic effects. A secondary screen is performed for all FDA-approved NSAIDs that reduce A(342 levels in cell cultures.
In a tertiary screen, A(3 production, sAPP production, and toxicity in a human neuroglioma cell line that expressed APP are examined for all FDA-approved NSAIDs and novel NSAIDs that selectively reduce A(342 levels. Three doses of each NSAID are tested. The first is a dose that is expected to cause maximum reduction of A[34a levels.
The second dose is one that reduces A(34z levels by 50% of the maximum value, while the third dose is one that reduces A(342 levels by 10-20% of the maximum value.
Tertiary screens are performed on the most potent NSAms identified by secondary screens.
NSAID toxicity is measured using an MTS assay (see Example 1) and a lactate dehydrogenase (LDH) release assay (Promega Corp, Madison, WI)e Example 19 -Acute sin,~le-dose studies to identify NSAIDs having in vivo activity To determine whether NSAIDs that selectively reduce SDS-soluble A(342 levels in cell culture studies also reduce brain A(342 levels, in vivo studies using Tg2576 mice are performed.
Mice of either sex are used for acute studies. Each experimental group, however, is performed using mice of the same sex. Power calculations, based on past measurements of variability of Tg2576 brain A(3 levels, indicate that an "n"
of five mice per study group gives an 80% chance of detecting a difference of 20% or more at p<0.05.
These calculations are supported by experiments on wortmanin treated and A(34a immunized Tg2576 mice, in which significant changes in A(3 levels, even between groups of three to four mice, were noted (Haugabook et al. (2000) Faseb .~. Although in most studies there are five mice per experimental group, in some instances, additional mice are used to account for loss due to death or illness. The use of additional mice also increases the power of ancillary studies such as those involving behavior, as sometimes, the number of mice needed to obtain a useful result is not known.

NSAIDs are prepared and administered to three-month-old Tg2576 mice as described in Example 13. To avoid extensive testing of NSATDs that are not active iu vivo, high doses of NSAIDs are used initially. NSAIDs are administered every four to eight hours; exact doses and dose schedules are determined from LDso values, half lives, and ih vitro dose response studies. In general, a maximum dose that is non-toxic, typically ranging from 1/10 tol/5 of the LDso value of the NSA)D, is used. If the LDSo and other phannacokinetic data of a given NSAID are unknown, their values are estimated using those of the nearest structural analogue.
To monitor toxicity, weights of a mouse before and after the study are compared.
In addition, one mouse from each treatment group is subjected to a liver function test (LFT) in which blood levels of two liver enzymes, SGOT and SGPT, are determined.
SGOT and SGPT are sensitive markers of liver toxicity. Furthermore, renal function, indicated by blood urea nitrogen (BLJN) levels, is determined. Tests for liver and renal functions are performed by Anilitics (Gaithersburg, MD), a company that specializes in these tests. Those NSAIDs having toxic effects at high doses are not used in long-term studies unless their effectiveness and lack of toxicity at lower doses are established.
Following a three-day administration schedule, mice are sacrificed; A(3 levels in plasma, brain, and CSF are determined; levels of NSAIDs in plasma are determined; and mice are examined for signs of toxicity. NSAIDs that selectively reduce SDS-soluble A(34a levels by more than 20-30% are examined in multiple dose response studies.
Example 20-Multiple-dose studies to identify doses useful for in vivo long-term ahimal ahd human studies NSAIDs that reduce A(34a levels in vivo, at high doses, are administered to groups of three mice at high, medium, and low doses using the same dosing regimen described in Example 19. A high dose is the amount used in the single dose screen of Example 19, while medium and low doses are determined by inference from ih vitro dose response studies described in Example 18. Those NSAIDs more potent than ibuprofen in vitro, (i.e., those that have ICso values required for maximum reduction of A[342 levels that are 30~ less than a mid ~.M value) are examined over a wide range of doses. For example, doses representing 1/50 to 1/10 of the ICso value are used in the multiple dose analysis.
NSAIDs having similar in vitro ICso values to ibuprofen are tested over a more limited range. For example, doses representing 1/10 to 1/3 of the ICSO value are used in the multiple dose analysis. Analyses of A(3 are performed as described for single dose studies. To identify plasma NSAID levels that correlate with A[342 reduction in vivo, a plasma NSAID level is determined for each dose examined using the HPLC method described in reference 64 and adapted for each particular NSAID. Data pertaining to plasma NSAID levels in these multiple dose studies are used as reference values for both long-term animal studies where NSAIDs are administered in feed, as well as for subsequent human studies.
Example 21- E~'ects of NSAIDs oh in vivo COX activity To determine if concentrations of NSAIDs used are sufficient to mediate anti-inflammatory effects, novel NSAIDs are examined for their in vivo COX
inhibitory activities and anti-inflammatory activities. For this study, the carrageneenan-induced footpad edema assay, described in Kalgutkar et al. (2000) J of Med Chem 43:2860-70, is performed on mice prior to sacrifice. For NSAIDs that do not reduce A(342 levels, the assays are performed on mice treated with NSAIDs at levels equivalent to that administered in long-term studies.
Example 22 - NSAIDs used in long-term preventative and therapeutic studies To determine whether the effects of NSAIDs on amyloid deposition in an animal model are attributable to direct inhibition of A(34a accumulation, or reduction in inflammatory processes in the brain, or both, the following groups of NSAIDs are examined in long-term preventative and therapeutic tests. NSAIDs that selectively reduce A(342 levels but lack anti-inflammatory properties, NSAIDs that selectively reduce A[342 levels and have anti-inflammatory properties, or NSAIDs that have no effect on A~i42 levels but have anti-inflammatory properties are examined in both preventative and therapeutic studies. Ibuprofen is used to examine indirect inflammatory-mediated effects on A(3 deposition and direct effects caused by reduction of A(34z levels, since it reduces A(342 levels and has anti-inflammatory properties. Celecoxib and naproxen, non-selective and selective Cox inhibitors, respectively, that do not cause reduction of A(342 levels are used to examine A[342-independent inflammatory-mediated effects. NSAIDs examined in both preventative and therapeutic studies include those that exhibit one of these three properties: selectivity for A(342 reduction relative to COX inhibition, A(34a reduction and COX-2 selectivity, or solely increased-potency for A[342 reduction in vivo.
Example 23 - Long-term NSAID dosing for preventative and therapeutic trials Long-term dosing of mice is achieved through feed. Feed containing the desired concentration of NSAID can be obtained from commercial entities. Prior to long-term preventative or therapeutic studies, successful administration of a chosen dose of NSAID
through feed is verified using the following experiment. First, an NSAID
concentration effective in reducing A(342 levels in acute studies, when administered by dropper, is chosen. This concentration corresponds to the lowest dose that can generate a maximum reduction in A[342 levels. In the case of an NSAID that does not reduce A(342 levels, a concentration sufficient to cause anti-inflammatory effects is chosen. In the case of ibuprofen, the dose that reduces A(342 levels also is a dose that causes anti-inflammatory effects. Feed containing the chosen concentration of NSAID is used in a short-term trial to compare mice given NSAID by dropper to mice given NSAID incorporated into feed.
The reduction in A(342 levels as well as peak plasma levels of NSAID are determined for mice given NSAID by dropper and mice given NSAID through feed. If levels of A(34~
reduction and peak plasma levels of NSAID in the two groups are comparable, then the chosen amount of NSAID is achieved through feeding, and long-term preventative or therapeutic studies are performed. If levels of A(342 reduction and peak plasma levels of NSAID in the two groups are not comparable, then the concentrations of NSAID
in feeds are altered appropriately until reduction in A(342 levels and peak plasma levels of NSAID
in the two groups of mice are comparable.

Example 24 - Determination of peak plasma levels o~NSAIDs Techniques for determination of ibuprofen, fenoprofen, and meclofenamic acid levels in plasma are described in Canaparo et al. (2000) Biomedical Chromatography 14:219-26; and Koup et al. (1990) Biopharmaceutics & Drug Disposition 11:1-15.
In general, an internal standard is added to a plasma sample. The sample is acidified and subjected to organic solvent extraction. The organic phase is dried, dissolved in a small volume, and subjected to HPLC using a C18 column. Calibration and standardization are carried out using untreated plasma spiked with NSAm for construction of a calibration curve.
10~
Example 25 - CSF collection Mice are anesthetized with pentobarbital (30-50 mg/kg). An incision from the top of the skull to the mid-back is made and the musculature from the base of the skull to the first vertebrae is removed to expose the meninges overlying the cisterns magna. The animal is placed on a narrow platform in an inverted fashion beneath a dissecting microscope. The tissue above the cisterns magna is excised with care not to puncture the translucent meninges. The surrounding area is cleaned gently with the use of cotton swabs to remove any residual blood or other interstitial fluid. The dilated cisterns magna containing CSF is easily visible at this point. In addition to the cerebellum, brain stem, and spinal cord, an extensive vascular network also is visible. A micro needle and a polypropylene narrow bore pipette are aligned just above the meninges. With care not to disrupt any of the underlying vasculature, the micro needle is slowly inserted into the cistern. The CSF, which is under a positive pressure due to blood pressure, respiration, and positioning of the animal, begins to flow out of the needle entry site once the micro needle is removed. The micro needle then is pulled slowly backwards and the narrow bore pipette is used to collect the CSF as it exits the compartment. Once the needle is completely removed, the pipette is lowered into the puncture site and used to remove any remaining CSF. The primary collection usually takes less than 15 seconds for completion. The cistern will refill with several p,L of CSF within two minutes. A second collection is performed to increase the net yield. At the end of the procedure, the emptied cistern is collapsed due to the removal of CSF. CSF is not collected past the first two minutes. The isolated CSF is transferred quickly into a pre-chilled polypropylene tube on ice. Less than 5% of samples contain visible blood contamination.
Example 26 - Biochemical, histochemical, behavioral, and toxicology evaluations of long-term NSAID treatment When mice are sacrificed, one hemi-brain is processed for biochemical analyses and the other for immunohistochemical and histochemical analyses.
A(~ao~ A(~aa~ and total A(3 levels in mice brains are determined. Both SDS-soluble and SDS-insoluble formic acid-soluble fractions are examined. ELISA, described in Kawarabayashi et al. (2001) J. Neur 21:372-381, and the BAN50 system, described in Suzuki et al. (1994) Sci 264: 1336-1340, are used. Both A(34o and A(34~
polyclonal capture antibodies and end-specific polyclonal antibodies are available.
Changes in levels of different A~i species due to NSAID treatments are examined by imunoprecipitation-mass spectral analysis. A(3 levels in plasma and CSF are determined at the time of sacrifice.
To examine total plaque burden, brain sections are stained with anti-A(3 antibodies. Antibodies to all A(3 species as well as end specific A(34o and A[3ø2 antibodies are used. Cored plaques are detected by staining with thioflavin. Plaque number and amyloid burden are calculated as described in the Sigma ScanPro image analysis software (see Haugabook et al. (2000) Faseb J). Plaque types and extent of vascular and parenchyrnal amyloid depositions are examined.
Inflammation is examined by biochemical and histochemcial techniques.
Astrocytosis is examined using immunohistochemical staining and Western blotting of the SDS-extract for GFAP. Microglial activation is examined using staining techniques for anti-phophotyrosine as described in Lim et al. (2000) JNeurosci 20:5709-14.
Alternatively, microglia are immunostained using a pan MHC antibody or using GS lectin as described in Frautschy et al. (1998) Am JofPath 152:307-17.
Inflammatory markers such as alACT and APOE are examined using Western blot analysis of the SDS-extract, while IL-1 and IL-6 are examined using commercially available ELISA kits.
To examine neuronal loss and tau pathology, sections from brains are stained using haematoxylon and eosin. Sections are examined for overt pathological signs and neuronal loss. Marked neuronal loss is quantitated using stereological counting. Tau pathology is assessed using immunohistochemical staining by several anti-phosphorylated tau antibodies.
For behavioral studies, a modified version of the Morris watermaze is used to S detect learning and memory impairments related to amyloidosis in mice over-expressing APP (see Chen et al. (2000) NatuYe 408:975-979). Testing is conducted in fully counterbalanced, age-matched squads of mice (five to seven per group); trial blocks are run at the same time each day, during the light cycle. Subjects run in a fixed order each day with an inter-trial interval of approximately fifteen minutes. Trial spacing minimizes effects of hypothermia and fatigue that often are seen in older animals (see Rick et al.
(1996) J Gerontol A Biol Sci Med Sci S 1:B2S3-60). The first day of testing consists of swimming to a visible platform. This assesses motivation, and visual and swimming ability. One trial is performed from a fixed starting position to each of four separate cued platform locations. In subsequent days, up to ten trials per day are performed using a 1S learning criterion of three consecutive trials with less than twenty escape latency (see Chen et al. (2000) Nature 408:975-979}. No probe trial is necessary since the only dependent variable measured is trials to reach criterion (TTC). Once an animal reaches criterion on one platform location, it is immediately switched to a new location. Testing is continued until five platform locations have been learned. Deficits in TTC are apparent in this paradigm primarily on the last two platform locations. These data are used with neuropathological data to assess the mice (see Chen et al. (2000) NatuYe 408:975-979).
Evaluation of neurological and sensorimotor skills is performed on the first day of testing, before the cued platform trial. A standard test battery is administered. This consists of (a) ten minutes in an automated open field, (b) examination of righting and 2S grasping reflexes, (c) latency to fall when suspended from a wire by the forepaws, and (d) rotorod performance. These tests screen basic functions such as strength, balance, and locomotor/exploratory behavior that can affect watermaze performance (Rick et al.
(1996) J Gerontol A Biol Sci Med Sci S 1: B2S3-60; Murphy et al. (1995) Neur Learn Mem 64:181-6; Bickford et al. (1997) Neur Aging 18, 309-18; Cammisuli et al.
(1997) Behav Brain Res 89:179-90; and Lewis et al. (2000) Nat Genet 25:402-S). In this way, effects of strength, balance, and locomotor/exploratory behavior on watermaze performance are accounted for.

As in acute studies, appropriate plasma markers are tested intermittently on a few NSAID treated mice to monitor liver and renal functions in both preventative and therapeutic trails. Weights of the mice are monitored bi-weekly, and complete blood counts are performed every two to three months. At the time of sacrifice, the GI tract is examined for signs of ulceration using a dissecting microscope as described in Kalgutkar et al. (2000) J of Med Chem 43:2860-70.
Example 27 - Determination of the effects of NSAIDs oh A,a deposition - long-team preventative trial NSAIDs that selectively reduce A[342 levels in acute studies are examined in a preventative trial to determine if they can prevent A(3 deposits. Six-month-old Tg2576 mice are used in preventative trials since A(3 deposit has not yet taken place. NSAID
treatment of mice at this age corresponds to treating humans before signs of clinical disease occur.
Tg2576 mice are treated with experimentally optimized doses of NSAID for three, six, and twelve months. Each treatment group consists of a minimum of twenty animals, five of which are examined at each of the three time points. The remaining five mice are included in case of illness or death during long-term dosing. Three to four mice are placed into a treatment group each month until groups of twenty animals are established.
At the time of sacrifice, tissues obtained for analysis are stored until all the mice within an experimental group have been sacrificed. Therefore, all samples from mice within one experimental group are examined simultaneously. For ibuprofen, naproxen, and control groups, twenty-seven mice are used per experimental group. The extra mice are treated for twelve months after which time behavioral patterns and additional pathologic parameters are examined.
The following NSAIDs are used in preventative trials: ibuprofen which reduces A~i4a levels, has anti-inflammatory activity, and has a short-half life;
meclofenamic acid which is more potent at reducing A(342 levels in vitro, and has anti-inflammatory activity;
sulindac which reduces A(342 levels, has anti-inflammatory activity, and has an extended-half life; naproxen which has no effect on A(342 levels, but has anti-inflammatory activity, and COX-1 and COX-2 inhibitory activities; and celecoxib which is an anti-inflammatory COX-2 selective agent. In addition, other NSAIDs that reduce A[342 levels but show selectivity for this effect over inhibitory effects on COX-1, COX-2, or both are included in this study. At three, six, and twelve months of treatment with NSAIDs, mice are analyzed for behavioral alterations; then they are sacrificed and biochemical analyses are performed as described in Example 26.
Example 28 - Alteration of A~i deposits by NSAIDs - long term therapeutic trial To determine whether A(3 deposition, the effects of A(3 deposition, or both can be altered once Aj3 has accumulated to a high level, NSAIDs that selectively reduce A(342 levels in acute studies are examined in a therapeutic trial. Effects of treatment with NSAIDs that reduce A(342 levels are compared to effects of treatment with NSAIDs that do not reduce A(342 levels such as the non-selective COX inhibitor naproxen and the selective COX inhibitor celecoxib. Sixteen-month-old Tg2576 mice are treated with experimentally optimized doses of NSAIDs for three or six months. Sixteen-month-old mice have large amounts of A(3 in the brain and therefore, are representative of human patients with clinical signs of AD. Amyloid deposition, behavior, and AD-like pathology are examined as described in Example 26. Fourteen mice per treatment group are used; at least five treated and five control mice are compared.
Example 29 - Statistical Analysis Mann-Whitney and Dunnet's tests are used for comparisons between groups of treated and untreated mice. A number of correlative comparisons are made.
Variables and outcomes used in statistical analysis for each study are the following.
For in vitro screening experiments, variables include: A[3 levels in media, NSAID
concentrations, toxicity, and COX inhibitory activity; while primary outcomes include reduction in A(34a levels and COX inhibitory activity. In acute single-dose studies, variables include: A[3 levels in brain, plasma, and CSF; NSAID concentrations in plasma and brain;
and dose of NSAID. Primary outcomes of acute single-dose studies include reduction in brain A(34a levels and plasma levels of NSAI17, while secondary outcomes includes correlation of brain, CSF, and plasma A(3 levels. In long-term studies, variables include:
A(3 levels in brain, plasma, and CSF; NSAID concentrations in plasma (and brain, if possible); dose of NSAID; amyloid burden; extent of inflammatory response; behavioral performance; and toxicity. Primary outcomes of long-term studies include effects on A~3 levels in the brain, while secondary outcomes include evaluation of inflammatory response, behavior, toxicity, and correlative analyses.
Example 30 - Clinical investi,~atiohs in amvloid reducin.~ actions ofNSAIDs The most promising FDA-approved NSAIDs, determined by preclinical studies, are examined for amyloid reducing actions in healthy subj ects as well as subj ects with mild to moderate Alzheimer's disease (AD). These studies are performed in three-group parallel design; each group consists of twelve subjects. Subjects are treated with an NSAID or a matching placebo several times a day, depending on the NSIAD, for fourteen days. Study NSAIDs are purchased and over-encapsulated by the San Diego VAMC Pharmacy service or by another compounding pharmacy. Placeboes are similarly encapsulated.
AD subjects are selected based on the following criteria. Subjects consist of men and women, ages 60-85, who are diagnosed with probable AD using the National Institute of Neurologic Communicative Disorders and Stroke-Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA) test (McKhann et al. (1984) Neurology 34:939-944) or have mild to moderate dementia as determined by the Mini-Mental State Examination (MMSE, Mohs et al. (1996) Int Psychogeriatr 8:195-203). MMSE
scores in the range of 15-25 indicate mild to moderate dementia. AD subjects have caregivers that can ensure compliance with medication regimens and with study visits and procedures.
Non-demented control subjects consist of men and women ages 60-80. Control subjects lack significant cognitive or functional complaints, or depression as determined by the Geriatric Depression Scale (GDS), and have MMSE scores in the range of 27-30.
Control subjects have the same general requirements as AD subjects with the exception that caregivers are not required. Both AD subjects and control subjects have good general health, i.e., subjects do not have serious or life-threatening comorbid conditions.
Subjects who have medically active major inflammatory comorbid conditions) such as rheumatoid arthritis, or those who have peptic ulcer, gastro-intestinal bleeding, or intolerance of NSAIDs in the past are excluded from the study. Those who have contra-indications to lumbar puncture, such as severe lumbar spine degeneration, sepsis in the region of the lumbar spine, or a bleeding disorder are excluded from participation in the study. In addition, subjects who currently or recently use medications such as NSAIDs, prednisone, or immunosuppressive medications such as cyclophosphamide that could interfere with the study are excluded. Recently is defined as within one month before undergoing the baseline visit (see next paragraph). Subjects undergoing acetylcholinsterase inhibitor (AChE-I) treatments for AD are not excluded if these subjects have been on stable doses for at least four weeks. Similarly, AD
subjects taking antioxidants such as vitamin E, vitamin C, or Gingko biloba are not excluded if they have been on stable doses for at least four weeks. Subjects who use NSAIDs or aspirin on a regular basis are excluded. If needed, analgesics such as paracetamol (Tylenol) are provided during the fourteen-day study.
The study procedure consists of three in-clinic visits: an initial screening visit, a baseline visit, and a follow-up visit at fourteen days. During the screening visit, information needed to assess eligibility is obtained and MMSE is administered.
During the baseline visit, which takes place within two weeks of the screening visit, physical examinations and lumbar punctures are performed. Blood samples are drawn for laboratory tests such as APO-E genotyping and for plasma preparation (see Example 31). At this time, subjects or caregivers, in the case of AD subjects, are given a fourteen-day supply of study NSAID along with instructions about timing of doses and potential adverse effects. (For AD subjects, caregivers are required to accompany subjects to each visit, and are responsible for monitoring and supervising administration of study NSAIDs.) A calendar is provided on which times of medications and potential adverse symptoms are recorded.
The NSAID treatment regimen consists of a fourteen-day treatment with NSAIDs in the form of capsules taken two or three times a day with meals. A
high and a low study dose of NSAID are used. For ibuprofen, study doses of 800 mg and 400 mg are used. A study dose of 800 mg consists of two 400 mg ibuprofen tablets, while a study dose of 400 mg consists of one 400 mg ibuprofen capsule and one placebo capsule. For sulindac, a study dose of 200 mg twice a day for a total of 400 mg per day is used. For meclofenamic acid, study doses of 100 mg and 400 mg per day are tested.
NSAIDs are pre-packed into a day-by-day plastic medication dispenser.

During the follow-up visit, twelve or fourteen days after beginning treatment, vital signs and adverse side effects of study NSAIDs are assessed. Surplus NSAIDs are returned and counted. In addition, lumbar punctures are performed and blood samples are drawn for laboratory tests and for plasma preparations.
Visits during which lumbar punctures are performed and blood samples are drawn are scheduled for mornings with overnight fasting to avoid obtaining post-prandial or hyperlipemic plasma samples, which can influence levels of A(34o and A(34z.
Table 4 summarizes biological markers that are analyzed from plasma and CSF
samples.
Table 4. Plasma and CSF biological markers Assay Method Volume of CSF Volume of Plasma Protein, glucose, 1 mL
cells A(34o ELISA 100 ~,L x2 (in 100 p.L
duplicate) x2 A(34a ELISA 100 ~,L x2 (in 100 ~.L
duplicate) x2 A[33g Mass Spectrometry1 mL

Isoprostanes Gas Chromatography/2 mL
Mass Spectrometry M-CSF ELISA SO pL x2 (in duplicate) MCP-1 ELISA 50 ~.L x2 (in duplicate) Tau, ELISA 50 p,L x2 (in duplicate) P-tau181 50 ~,L x2 (in duplicate) Plasma levels of HPLC 1 mL
NSAIDs ~

Example 31 - Collection of plasma and CSF
Plasma samples are prepared within 15-30 minutes after blood samples are drawn. Plasma samples are frozen at - 70 °C until used. At least 6 mL
of CSF and, whenever possible, 10-15 mL are drawn from each subject. Total cell, protein, and glucose estimations are performed. Samples are identified by a study ID
number, and technicians who run ELISAs or other assays are blinded to the identity of the subjects or the treatment conditions.
Example 32 - Specific assays ELISA is used to determine A(34o and A(342 levels in CSF. Batches of samples are assayed simultaneously in duplicate on microplates according to established procedures (3). In A(34a detection, two antibodies are used: (1) a monoclonal antibody that recognizes an epitope within the first five amino acids of A(3 is used for capture and (2) an end-specific monoclonal antibody that recognizes A~3 ending at amino acid 42 and conjugated to horse radish peroxidase is used for detection. CSF levels of A(33g are measured by mass spectroscopy as described in Example 6. CSF isoprostanes are measured by gas-chromatography/negative chemical ionization mass spectroscopy using internal standards for calibration Montine et al. (1999) Neurology 52:562-565). CSF
. levels of MCSF, MCP-1, tau, and P-tau181 are determined. Commercially available ELISA kits are used for M-CSF (R&D Diagnostics) and MCP-1 (Pharmingen, San Diego) determinations. CSF tau and P-tau181 are determined using ELISA kits from Innogenetics, Inc., Plasma levels of specific NSAIDs are determined by HPLC
methods described in published procedures (Canaparo et al. (2000) Biomed Chromatogr14:219-26).
Example 33 -Analysis of clinical data Reduction in A[34a levels due to NSAIDs treatment is detected as decreases in A[342 levels in CSF and/or plasma. Therefore, subjects with AD or elderly control subjects who receive NSAID treatments show serial decreases in CSF and/or plasma A(34a levels, while those who take a placebo will not show serial changes in CSF
and/or plasma A(342 levels.
To assess comparability between groups of subjects at baseline, demographic data (e.g. age and gender), dementia severity (MNISE score), and APO-E e4 allele frequency are compared between placebo groups, and groups of subjects with AD
or elderly controls that are treated with NSAIDs. Continuous variables are compared by ANOVA and frequencies of categorical variables such as gender and APO-E
genotype are compared using Chi-squared or Fisher's exact test.
Changes in levels of biomarkers of interest between baseline samples to follow-up samples are calculated for each subject. Descriptive statistics are used to determine whether levels of biomarkers at baseline are normally distributed. If they are, then mean changes in each treatment group are compared with each placebo group using ANOVA.
If they are not normal, then data transformation is applied or non-parametric statistics are used to compare changes in biomarker levels between different groups of subjects.

To determine whether changes in A(34a levels are accompanied by changes in A(34o and A(338 levels, CSF A[3 levels in placebo groups are compared to that in treatment groups using ANOVA. Levels of biomarkers related to microglial function (e.g.
M-CSF
and MCP-1), oxidative damage in the brain (e.g. F-2 isoprostanes), and neuronal degeneration (e.g. tau and P-taul8l) are compared before and after treatment as well as between groups treated with placebo or with NSAID. If levels of biological markers change after treatment with NSAID, the change is examined in relation to variables such as age, gender, APO-E genotype, and plasma NSAff~ levels. Scatter-plots and appropriate statistical comparisons are used.
Example 34 - Statistical Power Calculations Published data indicate that CSF A~i42 levels remain stable on repeated lumbar punctures. The power to detect differences between subjects treated with NSAIDs and subjects treated with placeboes depends on magnitudes of changes in biomarker levels after treatment relative to baseline.
In published longitudinal data for CSF A[342 levels in an AD patient group of (see Andreasen et al. (1999) Arch Neurol 56:673-80), baseline CSF A(342 level (mean ~
SD) was 709 ~ 304 pg/ml and follow-up (10 months later) CSF A(34a level was 701 ~ 309 pg/mL. The correlation between the first and second CSF A(342 level was R =
0.90. No published longitudinal CSF A(342 data are available in healthy subjects. In two studies that included healthy subjects, the values for CSF A[342 levels were 1485 ~
473 pg/mL
(see Galasko et al. (1998) Arch Neurol 55:937-45) and 1678 ~ 436 pg/mL (see Andreasen et al. (1999) Arch Neurol 56:673-80).
The power calculation uses the following assumptions: (1) levels are stable over time as described in Andreasen et al. (1999) Arch Neurol 56:673-80 and (2) variance of change is similar. The standard deviation is calculated as square root of ((1-correlation)*2*SD~2). A pre-post correlation of 0.8 for CSF A[342 Ievel is assumed.
If the change in pre-post mean CSF A(3 levels is assumed to be approximately zero in the placebo group, then effect size depends on the mean level of A(342 at baseline.
For example, for elderly controls, if the mean CSF A~34z level is 1485 pg/mL
(see Galasko et al. (1998) Arch Neurol 55:937-45), then a 0.25 effect size represents an increase or decrease of the mean by 371 pg/mL due to treatment.
For power calculations, the following are assumed: (1) alpha = 0.05, (2) power =
0.80, and (3) two-group studies in which equal numbers of subjects exposed to placebo and treatment are used. For power calculations with an effect size of 0.25, a sample size (1~ of 11 in each of the two groups is required. With effect size of 0.2, an N
of 16 is required in each group.
Twelve subjects per group are used for each study allowing for detection of an effect size of 0.25 or higher. In pre-clinical studies, several NSAIDs (including ibuprofen and meclofenamic acid) reduced A(342 levels in supernatants from cultured cells and in brain tissues of transgenic mice by over 25%. In long-term transgenic mouse studies using ibuprofen, reported in (Lim et al. (2000) JNeurosci 20:5709-14), A(3 levels in the brain were about 38% lower when treated than untreated.
If the variance in CSF A[342 levels between subjects or on repeated lumbar puncture is greater than in these projections, then sample size is re-assessed and group size is modified as needed. A similar set of calculations using published data on CSF
Aj34~ levels in AD patients shows that groups of twelve patients are sufficient to detect a 25% effect size.
Published longitudinal CSF data are available for CSF tau in AD. Sunderland et al. (1999) Biol Psychiatry 46:750-755 studied twenty-nine patients with AD
having baseline CSF tau (mean ~ SD) of 548 ~ 355 pglmL, follow-up CSF tau at twelve months of S57 ~ 275 pg/mL, and an R-value of 0.85.
The decision to use twelve subjects per group is derived from A(3 data. Again, assuming CSF tau remains stable and unchanged on average in the absence of treatment, an effect size for a decrease in CSF tau by at least 33% relative to baseline is 183 pg/mL
of tau.
In a two-group study design with (1) equal subject numbers receiving placebo and treatment, (2) N =12 per group, and (3) assuming a = 0.05, then power is 73%
for detecting an effect size of 33% or greater for tau.
With the exception of plasma A(3 levels that remained stable as indicated by preliminary ibuprofen studies, the degree of variation of longitudinal measurements of other biomarkers is not known. Ibuprofen studies in healthy elderly and subjects with mild AD are performed first, then sample sizes are reassessed for all biomarkers measured and necessary changes are incorporated in to other NSAID studies.
Example 35 - Placebo-controlled study of NSAIDs with A,~3-lowering actions A double-blind randomized placebo-controlled study is performed using sixty AD subjects treated with a placebo, ibuprofen, or another FDA-approved NSAID
with A[3-reducing action at a well-tolerated dose for 48 weeks. Specific NSAIDs and doses are selected based on results obtained in Example 30.
Subjects are 50-90 years of age and have diagnoses of probable AD as indicated by the NINCDS-ADRDA test. Subjects have an MMSE range of 15-25, good general health, i.e., no life threatening or major medical illnesses; and caregivers who can supervise medication regimens and provide collateral information. Additional screening criteria are as described in Example 30.
Initially, subjects are assessed for eligibility in a screening visit. MMSEs and physical examinations are performed. Blood samples are obtained for routine laboratory tests. Block randomization is used to assign patients to placebo or active treatment groups. Assignment is determined according to baseline MMSE scores so that dementia severity is similar in the placebo and active treatment groups.
During the baseline visit, scheduled within two weeks of the screening visit, vital signs are assessed, lumbar punctures are performed, and blood samples are drawn for APO-E genotyping and for plasma preparation (see Example 31). CSF levels of A(34oa A(34a, isoprostanes, tau, and P-tau as well as plasma levels of A/34o and A(342 are determined. In addition, cognition is assessed using the Alzheimer's Disease Assessment Scale - cognitive component (ADAS-cog, see Galasko et al. (1997) Alzheimer Dis Assoc Disord 11; Suppl 2:533-9) and MMSE, while functional ability is assessed using the Alzheimer's Disease Cooperative Study Activities of Daily Living Scale (ADCS-ADL) (see McKhann et al. (1984) Neurology 34:939-944). At this time, caregivers of subjects are given a twelve-week supply of study NSAID along with instructions on timing of doses and potential adverse effects.
At the 12-week visit, vital signs, stool guaiac, and adverse side effects are assessed. Unused NSAID is counted. At the 24-week visit, assessment procedures identical to those of the baseline visit are performed. A count of unused NSAIDs and an inquiry about adverse events are made. At the 36-week visit, assessment procedures identical to the 12-week visit are performed, while at the 48-week visit, assessment procedures identical to those of the baseline visit are performed. A count of unused NSAIDs and inquiry about adverse events are made. Table 5 summarizes the examinations performed at each visit in the study.
Table 5. Schedule of events ScreenBaseline12 week24 week 36 week48 week Check entry criteria,X
obtain consent Screening blood testsX

Demographics, medicalX
history Vital signs X X X X X X

Rectal examination, X X X X X
stool guaic MMSE X X X X

ADAS-cog, ADCS ADL X X X

Dispense medications X X X

Adverse events, pill X X X X
count Lumbar puncture, X X
plasma for A(3 Blood drawn for safety X X X
laboratory tests In addition, each subject/caregiver is interviewed by telephone at 4, 8, 16, and 20 weeks to inquire about continuation in the study, medication usage, and adverse events.
Example 36 - Statistical analyses of placebo-controlled studies Statistical analyses involve the comparison of cognitive (ADAS-cog, MMSE), functional (ADCS-ADL), and biomarker data of subjects before and after treatment.
Subjects treated with NSAID for 48 weeks are expected to exhibit less cognitive and functional decline relative to subjects who are treated with placebo. NSAID
treatments are expected to associate with improved biomarker indices in CSF and possibly in plasma.
Differences (~s) between final and initial ADAS-cog and ADCS-ADL scores are referred to as primary outcome measuxes. Mean Os for placebo and treatment groups are compared by ANOVA. To control for subjects who fail to complete the study, a Last Observation Carried Forward (LOCF) analysis is performed.
Changes in CSF levels of A(34a, tau, P-tau181, F-2-isoprostanes, and plasma A~i4z and A/34o are similarly analyzed as outcome measures using ANOVA, or a non-parametric test (e.g. Kruskal-Wallis) if the data are not normal. Correlations between changes in biomarker measures and in clinical measures at 24 weeks are examined by scatter-plots and correlational analyses.
OTHER EMBODIMENTS
It is to be understood that while the invention has been described in conjunction with the detailed description thereof, the foregoing description is intended to illustrate and not limit the scope of the invention, which is defined by the scope of the appended claims.
Other aspects, advantages, and modifications are within the scope of the following claims.

Claims (54)

WHAT IS CLAIMED IS:

1. A method of preventing, delaying, or reversing the progression of Alzheimer's disease, said method comprising:
(a) identifying a mammal in need of prevention, delay, or reversal of the progression of Alzheimer's disease, (b) administering an A.beta.42 lowering agent to said mammal under conditions in which A.beta.42 levels are selectively reduced.

2. The method of claim 1, wherein the level of A.beta.38 is increased.

3. The method of claim 1, wherein levels of one or more of A.beta.34, A.beta.36, A.beta.37, and A.beta.39 are increased.

4. The method of claim 1, wherein the level of A.beta.40 is unchanged.

5. The method of claim 1, wherein said A.beta.42 lowering agent is an aryl propionic acid derivative, an aryl acetic acid derivative, or an amino carboxylic acid derivative.

6. The method of claim 1, wherein said A.beta.42 lowering agent is a structural derivative of an NSAID selected from the group consisting of flufenmic acid, meclofenamic acid, fenoprofen, carprofen, ibuprofen, ketoprofen, and flurbiprofen.

7. The method of claim 1, wherein said A.beta.42 lowering agent is a structural derivative of 5-nitro-2-(3-phenylpropylamino)benzoic acid).

8. The method of claim 1, wherein said A.beta.42 lowering agent lacks COX-1, COX-2, or both COX-1 and COX-2 inhibiting activity.

9. The method of claim 1, wherein said A.beta.42 lowering agent has a greater potency for lowering A.beta.42 levels than for inhibiting COX-1, COX-2, or both COX-1 and COX-2 activity.

10. The method of claim 1, wherein said mammal is a human.

11. The method of claim 1, wherein said mammal has not been diagnosed with Alzheimer's disease.

12. The method of claim 1, wherein said mammal does not have a genetic predisposition for Alzheimer's disease.

13. A method for developing an A.beta.42 lowering agent, said method comprising:
(a) derivatizing the NSAID meclofenamic acid or flufenamic acid by altering the position of the carboxylic acid group on the phenyl ring of said NSAID, altering the position or type of substituents on the phenyl ring opposite the carboxylic acid group of said NSAID, altering the bond connecting the two phenyl rings of said NSAID, altering the carboxylic acid group of said NSAID to propionic acid or another substituent, or performing any combination of these alterations, to generate a candidate A.beta.42 lowering agent; and (b) determining the effects of said candidate A.beta.42 lowering agent on levels of A.beta.42 and A.beta.38 in a biological composition following contact of said candidate A.beta.42 lowering agent with said biological composition, wherein a decrease in the level of A.beta.42 together with an increase in the level of A.beta.38 indicate that said candidate A.beta.42 lowering agent is an A.beta.42 lowering agent.

14. A method for developing an A.beta.42 lowering agent, said method comprising:
(a) providing an NSAID selected from the group consisting of fenoprofen, flurbiprofen, and carprofen;
(b) altering the position of the propionic acid group on the phenyl ring of said NSAID, altering the position or type of substituents on the phenyl ring opposite the propionic acid group of said NSAID, altering the bond connecting the two phenyl rings of said NSAID, altering the acetic acid group of said NSAID to carboxylic acid or another substituent, or performing any combination of these alterations, to generate a candidate A.beta.42 lowering agent; and (c) determining the effects of said candidate A.beta.42 lowering agent on levels of A.beta.42 and A.beta.38 in a biological composition following contact of said candidate A.beta.42 lowering agent with said biological composition, wherein a decrease in the level of A.beta.42 together with an increase in the level of A.beta.38 indicate that said candidate A.beta.42 lowering agent is a novel A.beta.42 lowering agent.

15. A method for developing an A.beta.42 lowering agent, said method comprising:
(a) altering the carboxylic acid group of indomethacin to another substituent, altering the indole nitrogen to another substituent, or performing any combination of these alterations to generate a candidate A.beta.42 lowering agent; and (b) determining the effects of said candidate A.beta.42 lowering agent on levels of A.beta.42 and A.beta.38 in a biological composition following contact of said candidate A.beta.42 lowering agent with said biological composition, wherein a decrease in the level of A.beta.42 together with an increase in the level of A.beta.38 indicate that said candidate A.beta.42 lowering agent is said novel A.beta.42 lowering agent.

16. A method for developing an A.beta.42 lowering agent, said method comprising:
(a) altering the methylthiol group of sulindac sulfide to another substituent, altering the propionic acid group of sulindac sulfide to another substituent, altering the fluoride moiety of sulindac sulfide to another substituent, or performing any combination of these alterations, to generate a candidate A.beta.42 lowering agent; and (b) determining the effects of said candidate A.beta.42 lowering agent on levels of A.beta.42 and A.beta.38 in a biological composition following contact of said candidate A.beta.42 lowering agent with said biological composition, wherein a decrease in the level of A.beta.42 together with an increase in the level of A.beta.38 indicate that said candidate A.beta.42 lowering agent is a novel A.beta.42 lowering agent.

17. A method of identifying an A.beta.42 lowering agent useful for preventing, delaying, or reversing the progression of Alzheimer's disease, said method comprising:
(a) identifying a candidate A.beta.42 lowering agent;
(b) contacting said candidate A.beta.42 lowering agent with a biological composition comprising APP and an APP processing activity under conditions in which said APP processing activity occurs;
(c) comparing the level of A.beta.42 in said biological composition contacted with said candidate A.beta.42 lowering agent to the level of A.beta.42 in a biological composition not contacted with said candidate A.beta.42 lowering agent;
(d) identifying said candidate A.beta.42 lowering agent as an A.beta.42 lowering agent, useful for preventing, delaying, or reversing the progression of Alzheimer's disease, if a reduction in the level of A.beta.42 in said biological composition contacted with said candidate A.beta.42 lowering agent is observed when compared with the level of A.beta.42 in said biological composition not contacted with said candidate A.beta.42 lowering agent.

18. A method of identifying an A.beta.42 lowering agent useful for preventing, delaying, or reversing the progression of Alzheimer's disease, said method comprising:
(a) identifying a candidate A.beta.42 lowering agent;
(b) contacting said candidate A.beta.42 lowering agent with a biological composition comprising A.beta.42 and an A.beta.42 catabolic activity under conditions in which said A.beta.42 catabolism occurs;
(c) comparing the level of A.beta.42 in said biological composition contacted with said candidate A.beta.42 lowering agent to the level of A.beta.42 in a biological composition not contacted with said candidate A.beta.42 lowering agent;
(d) identifying said candidate A.beta.42 lowering agent as an A.beta.42 lowering agent, useful for preventing, delaying, or reversing the progression of Alzheimer's disease, if a reduction in the level of A.beta.42 in said biological composition contacted with said candidate A.beta.42 lowering agent is observed when compared with the level of A.beta.42 in said biological composition not contacted with said candidate A.beta.42 lowering agent.

19. A method for identifying a novel A.beta.42 lowering agent that has a greater potency for lowering A.beta.42 levels than for inhibiting COX-1, COX-2, or both COX-1 and COX-2 activity, said method comprising:
(a) identifying an A.beta.42 lowering agent by screening for the ability to lower the level of A.beta.42 in a biological composition;
(b) determining the IC50 of said A.beta.42 lowering agent for A.beta.42 lowering by performing dose response studies;
(c) determining whether said A.beta.42 lowering agent inhibits COX-1, COX-2, or both COX-1 and COX-2 using in vitro COX-1 and COX-2 inactivation assays;

(d) comparing said IC50 for A.beta.42 lowering to said IC50 for COX-1, COX-2, or both COX-1 and COX-2 inhibition, wherein an IC50 for A.beta.42 lowering that is greater than ten-fold the IC50 for COX-1, COX-2, or both COX-1 and COX-2 inhibition indicates that said A.beta.42 lowering agent is one that has greater potency for lowering A.beta.42 levels than for inhibiting COX-1, COX-2, or both COX-1 and COX-2 activity.

20. The method of claim 19, wherein said greater potency is further confirmed by demonstrating that administration of said A.beta.42 lowering agent to an animal reduces A.beta.42 levels at doses that do not inhibit or only minimally inhibit COX-1, COX-2, or both COX-1 and COX-2 activity to levels that do not cause significant clinical side-effects upon administration of said A.beta.42 lowering agent.

21. The method of claim 13, 14, 15, 16, 17, 18, or 19, wherein said biological composition comprises an enzyme.

22. The method of claim 13, 14, 15, 16, 17, 18, or 19, wherein said biological composition comprises a mammalian cell.

23. The method of claim 13, 14, 15, 16, 17, 18, or 19, wherein said biological composition comprises a transgenic animal.

24. The method of claim 13, 14, 15, 16, 17, 18, or 19, wherein the level of A.beta.40 is unchanged.

25. The method of claim 13, 14, 15, 16, 17, 18, or 19, wherein levels of one or more of A.beta.34, A.beta.36, A.beta.37, and A.beta.39 are increased.

26. The method of claim 17, 18, or 19 wherein the level of A.beta.38 is increased.

27. The method of claim 17, 18, or 19, wherein said candidate A.beta.42 lowering agent is selected from the group consisting of aryl propionic acid derivative, an aryl acetic acid derivative, and an amino carboxylic acid derivative.

28. The method of claim 17, 18, or 19, wherein said candidate A.beta.42 lowering agent is a structural derivative of an NSAID selected from the group consisting of flufenmic acid, meclofenamic acid, fenoprofen, carprofen, ibuprofen, ketoprofen, and flurbiprofen.

29. The method of claim 13, 14, 15, 16, 17, or 18, wherein said candidate A.beta.42 lowering agent lacks COX-1, COX-2, or both COX-1 and COX-2 inhibiting activity.

30. The method of claim 13, 14, 15, 16, 17, or 18, wherein said candidate A.beta.42 lowering agent has a much greater potency in vivo for lowering A[34a relative to COX-l, COX-2, or both COX-1 and COX-2 inhibiting activity.

31. A method of identifying an agent that increases the risk of developing, or hastens progression of, Alzheimer's disease in a patient, said method comprising:

(a) identifying a candidate agent;
(b) contacting said candidate agent with a biological composition comprising APP and an APP processing activity under conditions in which said APP
processing activity occurs;
(c) comparing the level of A.beta.42 in said biological composition contacted with said candidate agent to the level of A.beta.42 in a biological composition not contacted with said candidate agent;
(d) identifying said candidate agent as one that can increase the risk of developing, or hasten the progression of, Alzheimer's disease if an increase in the level of A[34a in said biological composition contacted with said agent is observed when compared with the level of A.beta.42 in said biological composition not contacted with said agent.

32. A method of identifying an agent that increases the risk of developing, or hastens the progression of, Alzheimer's disease in a patient, said method comprising:
(a) identifying a candidate agent;
(b) contacting said candidate agent with a biological composition comprising A.beta.42 and an A.beta.42 catabolic activity under conditions in which said A.beta.42 catabolism occurs;
(c) comparing the level of A.beta.42 in said biological composition contacted with said candidate agent to the level of A.beta.42 in a biological composition not contacted with said candidate agent;
(d) identifying said candidate agent as one that can increase the risk of developing, or hasten the progression of, Alzheimer's disease if an increase in the level of A.beta.42 in said biological composition contacted with said agent is observed when compared with the level of A.beta.42 in said biological composition not contacted with said agent.

33. The method of claims 31 and 32, wherein said biological composition comprises an enzyme.

34. The method of claims 31 and 32, wherein said biological composition comprises a mammalian cell.

35. The method of claims 31 and 32, wherein said biological composition comprises a transgenic animal.

36. A composition comprising an A.beta.42 lowering agent and an antioxidant.

37. The composition of claim 36, wherein said antioxidant is selected from the group consisting of vitamin E, vitamin C, curcumin, and Gingko biloba.

38. A composition comprising an A.beta.42 lowering agent and a non-selective secretase inhibitor.

39. A composition comprising an A.beta.42 lowering agent and an acetylcholinesterase inhibitor.

40. A kit comprising an A.beta.42 lowering agent and an antioxidant.

41. A kit comprising an A.beta.42 lowering agent and a non-selective secretase inhibitor.

42. A kit comprising an A.beta.42 lowering agent and an acetylcholinesterase inhibitor.

43. The kit of claim 40, 41, or 42, wherein said kit comprises instructions that indicate a dose regimen for use of said A.beta.42 lowering agent with said antioxidant, said secretase inhibitor, or said acetylcholinesterase inhibitor.

44. The use of an A.beta.42 lowering agent in the manufacture of a medicament for the treatment of Alzheimer's disease, wherein administration of said A.beta.42 lowering agent to a patient is effective for reducing A.beta.42 levels.

45. The use as in claim 44, wherein said agent is further effective for increasing A[338 levels.

46. The use as in claim 44, wherein said agent is further effective for increasing A[33a, A.beta.36, A.beta.37, or A.beta.39 levels.

47. The use as in claim 44, wherein the level of A.beta.40 is unchanged.

48. The use as in claim 44, wherein said A.beta.42 lowering agent is an aryl propionic acid derivative, an aryl acetic acid derivative, or an amino carboxylic acid derivative.

49. The use as in claim 44, wherein said wherein said A.beta.42 lowering agent lacks COX-1, COX-2, or both COX-1 and COX-2 inhibiting activity.

50. The use as in claim 44, wherein said wherein said A.beta.42 lowering agent has a greater potency in vivo for lowering A.beta.42 levels than for inhibiting COX-1, COX-2, or both COX-1 and COX-2 activity.

51. The use as in claim 44, wherein said Alzheimer's disease is in a mammal.

52. The use as in claim 44, wherein said mammal is a human.

53. The use as in claim 44, wherein said mammal has not been diagnosed with Alzheimer's disease.

54. The use as in claim 44, wherein said mammal does not have a genetic predisposition for Alzheimer's disease.