USF Libraries
USF Digital Collections

Effects of long-term administration of caffeine in a mouse model for Alzheimer's disease


Material Information

Effects of long-term administration of caffeine in a mouse model for Alzheimer's disease
Physical Description:
Schleif, William
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:


Subjects / Keywords:
Transgenic mice
Dissertations, Academic -- Biology -- Masters -- USF
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


ABSTRACT: A recent epidemiological study suggested that higher caffeine intake reduces the risk of Alzheimer's disease (AD). Caffeine, a widely consumed stimulatory drug, is a non-selective adenosine receptor antagonist that has been shown to increase plasma adenosine levels in rodents. To determine any long-term protective effects of caffeine in a controlled longitudinal study, caffeine was added to the drinking water of APPsw transgenic (Tg) mice between 4 and 9 1/2 months of age, with behavioral testing done during the last 6 weeks of treatment. The average daily intake of caffeine per mouse (1.5 mg) was the human equivalent of 5 cups of coffee/day. Across multiple cognitive tasks of spatial learning/reference memory, working memory, and recognition/identification, Tg mice given caffeine (Tg+Caff) performed significantly better than Tg control mice and similar to non-transgenic controls. Discriminant Function Analysis involving multiple cognitive measures clearly showed the^ superior overall cognitive performance of Tg+Caff mice compared to Tg controls. Analysis of amyloid beta in the hippocampus by ELISA revealed Tg+Caff mice had significantly less soluble amyloid beta 1-40 and insoluble amyloid beta 1-42. In a follow-up study involving neurochemical analysis only, caffeine was added to the drinking water of 17 month old APPsw mice for 18 days. In this study, Tg+Caff mice also showed a significant reduction of insoluble amyloid beta 1-42 in the hippocampus. In contrast to the reduced extracellular brain levels of adenosine in Tg controls, caffeine treatment normalized brain adenosine levels in Tg mice to that of non-transgenic controls. Analysis of amyloidogenic secretase activity revealed the reduction in amyloid beta is likely because of a reduction in gamma secretase activity as a result of increased SAM silencing of PS1 expression. This study suggest that a modest, long-term caffeine intake of approximately 500 mg per day (5 cups of coffee) may redu ce considerably the risk of AD by decreasing amyloidogenesis.
Thesis (M.A.)--University of South Florida, 2005.
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by William Schleif.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 133 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001912709
oclc - 174041030
usfldc doi - E14-SFE0001342
usfldc handle - e14.1342
System ID:

This item is only available as the following downloads:

Full Text


Effects of Long-Term Administration of Caffeine in a Mouse Model for Alzheimers Disease by William Schleif A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science Department of Biology College of Arts and Sciences University of South Florida Major Professor: Gary Arendash, Ph.D. Huntington Potter, Ph.D. Sidney Pierce, Ph.D. Date of Approval: September 12th, 2005 Keywords: amyloid, S-adenosylmethionine PS1, adenosine, transgenic mice Copyright 2005, William Schleif


Acknowledgements I thank Dr. Edward Jackson and his laboratory for their analysis of brain adenosine receptors and the determination of extracellular brain adenosine levels in the course of this study. I would also like to thank Dr. Jun Tan and Kavon Rezaizadeh for their work in measuring alphaand betaCTFs and their assistance in determining brain concentrati ons of SAM and PS1 activity.


i Table of Contents List of Tables .....................................................................................................................iii List of Figures ....................................................................................................................iv ABSTRACT I. Alzheimers Disease ........................................................................................................1 Behavior Characterization ......................................................................................1 Pathological Characterization ................................................................................2 Genetics of Alzheimers Disease .............................................................................5 Diagnosis of Alzheimers Disease ..........................................................................7 Risk Factors for Alzheimers Disease ...................................................................11 Treatments for AD .................................................................................................12 II. Animal Models of AD ..................................................................................................13 PDAPP Model .......................................................................................................14 APP sw and APP23 Models ....................................................................................16 PSAPP Model ........................................................................................................24 Limits of Animal Models for Alzheimers disease .................................................27 III. Caffeine Consumption and Alzheimers Disease .......................................................28 Pharmacological Profile of Caffeine ....................................................................29 Molecular Actions of Caffeine ..............................................................................30 Immediate Effects of Caffeine Intake ....................................................................38 Long-Term Effects of Caffeine Intake ...................................................................42 Tolerance to Caffeine ............................................................................................47 Health Risks of Caffeine Intake .............................................................................49 Summary ...............................................................................................................50 IV. Specific Aims .............................................................................................................52


ii V. Materials and Methods .................................................................................................54 Effects of Long-Term Caffeine Administration in Young Adult APP sw Mice ........54 Effects of Caffeine Administration in Aged APP sw Mice .......................................65 Determination of Soluble/Insoluble A Levels .....................................................69 Statistical Analysis ................................................................................................70 VI. Results ........................................................................................................................72 BehaviorSensorimotor Evaluation .....................................................................72 BehaviorCognitive Evaluation ...........................................................................73 Multi-metric Statistical Analysis ...........................................................................87 Neuropathologic/Neurochemical Measures: Study A ...........................................92 Neuropathologic/Neurochemical Measures: StudyB ............................................95 VII. Discussion ...............................................................................................................103 General Summary ...............................................................................................103 Proposed Mechanism of Caffeine -Mediated Cogniti ve Improvement ................112 Clinical Implications of Caffein e Administration Study Findings and Potential Future Investigations ...........................................................................115 References .......................................................................................................................117


iii List of Tables Table 1. Immediate effects of moderate caffeine intake 42 Table 2. Mouse adenosine recepto r PCR primers and cDNA sizes... 65 Table 3. Factor loadings of behavioral measures.. 88 Table 4. Summary of discrimi nant function analyses 90


iv List of Figures Figure 1. Receptor-mediated effects of caffeine intake.. 38 Figure 2. Impact of elevated levels of homocysteine and decreased SAM in Alzheimers disease... 45 Figure 3. Relationship between habitual caffeine intake and Alzheimers disease. 47 Figure 4. General protocol time line for long-term caffeine administration study. 55 Figure 5. General protocol time line for caffeine administration aged mice. 66 Figure 6. Open field, balance beam, and string agility performance for NT, Tg, and Tg+Caff mice. 74 Figure 7. Elevated plus-maze perfor mance for NT, Tg, and Tg+Caff mice.. 75 Figure 8. Y-maze performance for NT, Tg, and Tg+Caff mice.. 76 Figure 9. Morris maze acquisition in NT, Tg, and Tg+Caff mice.. 78 Figure 10. Morris maze retention in NT, Tg, and Tg+Caff mice.. 79 Figure 11. Circular platform escape latencies in NT, Tg, and Tg+Caff mice. 81 Figure 12. Platform recognition performance in NT, Tg, and Tg+Caff mice.. 83 Figure 13. RAWM performance in NT, Tg, and Tg+Caff mice... 84 Figure 14. Overall RAWM performance in NT, Tg, and Tg+Caff mice.. 86


v Figure 15. Canonical score plots of step wise-forward discriminant function analyses used to compared overall cognitive performance of NT, Tg+Caff, and Tg mice.. 91 Figure 16. Determination of adenosine A1 and A2A receptor densities in striatum, hippocampus, and fr ontal cortex in NT, Tg, and Tg+Caff mice... 93 Figure 17. Analysis of A 1-40 and 1-42 in Tg and Tg+Caff mice (study A)... 94 Figure 18. Comparison of -CTF to CTFs ratio in Tg and Tg+Caff mice... 96 Figure 19. Analysis of A 1-40 and 1-42 in Tg and Tg+Caff mice (study B) 97 Figure 20. GammaSecretase activ ity in cerebral cortex of APP sw mice following caffeine administration 99 Figure 21. SAM ( S-adenosyl-methionine) levels in cerebral cortex of APP sw mice following caffeine administration.....................................100 Figure 22. Assay of extracellula r brain adenosine levels.......................................102


vi Effects of Long-Term Administ ration of Caffeine in a Mouse Model for Alzheimers Disease William Schleif ABSTRACT A recent epidemiological study suggested that higher caffeine intake reduces the risk of Alzheimers disease (AD). Caffeine, a widely consumed stimulatory drug, is a non-selective adenosine receptor antagonist th at has been shown to increase plasma adenosine levels in rodents. To determine a ny long-term protective effects of caffeine in a controlled longitudinal study, caffeine was added to the drinking water of APP sw transgenic (Tg) mice between 4 and 9 months of age, with behavioral testing done during the last 6 weeks of treatment. The av erage daily intake of caffeine per mouse (1.5 mg) was the human equivalent of 5 cups of coffee/day. Across multiple cognitive tasks of spatial learning/reference memory, work ing memory, and recognition/identification, Tg mice given caffeine (Tg+Caff) performed si gnificantly better than Tg control mice and similar to non-transgenic controls. Discriminant Function Analysis involving multiple cognitive measures clearly showed the superior overall cognitive performance of Tg+Caff mice compared to Tg controls. Analysis of A in the hippocampus by ELISA revealed Tg+Caff mice had si gnificantly less soluble A 1-40 and insoluble A 1-42 In a follow-up study involving neurochemical analysis only, caffeine was added to the drinking water of 17 month old APP sw mice for 18 days. In this study, Tg+Caff mice also


vii showed a significant reduction of insoluble A 1-42 in the hippocampus. In contrast to the reduced extracellular brain levels of ade nosine in Tg controls, caffeine treatment normalized brain adenosine levels in Tg mice to that of non-tr ansgenic controls. Analysis of amyloidogenic secretase activ ity revealed the reduction in is likely because of a reduction in secretase activity as a result of increased SAM silencing of PS1 expression. This study suggest th at a modest, long-term caffein e intake of approximately 500 mg per day (5 cups of coffee) may redu ce considerably the ri sk of AD by decreasing amyloidogenesis.


1 I. Alzheimers Disease With exponential growth in the population of aged individuals in industrialized countries comes the increasing prevalence of age-related disorders. Foremost amongst these disorders is Alzheimers disease (AD), which is currently the leading cause for dementia in the elderly and afflicts an es timated 4.5 million people in the United States alone. This number is expected to triple over the next fifty years if a cure is not found (Hebert et al., 2003). The onset of Alzheimers is as tragic as it is devastating; progressively disrupting areas of the brain responsible for me mory and motor skills that ultimately leave the patient comatose and una ble to communicate or relate with their environment before they succumb to death. This continual decline of the patients lifestyle leaves them in a state that is a st rain on their family and is costly to society. Behavior Characterization The initial onset of AD is often times difficult to pinpoint During the latent period of the disease, the underlying disruptions in norma l brain physiology begin to occur decades before any cognitive decline is noticed. Some patients develop a condition termed MCI (Mild Cognitive Impairment), wh ich is characterized by the progressive decline of short-term memory and indicates a substantial risk for developing AD. An estimated 10-15% of MCI patients progress into AD per year, yet it is also worth mentioning that not all MCI subjects develop AD (Petersen et al., 2001). MCI represents a transitional state where patients have a noticeab le decline in memory different from that


2 of normal aging but are not yet diagnosabl e as having AD. The short-term memory impairment found in MCI is similar to AAMI (Age-Associated Memory Impairment), but cognitive testing can distinguish between the two. Some MCI patients exhibit neuropsychiatric symptoms as well. A recen t case study found over 30% of MCI patients exhibited symptoms of depression, aggression, anxiety, apathy and irritability. These symptoms follow a similar trend seen in early AD patients and may serve as a clinical indicator of MCI severity (Feldman et al., 2004). Early AD is marked by a moderate loss in working memory that is frequently accompanied by depression and the symptoms mentioned earlier. Typically language difficulties also appear with the patient struggling with tests that involve word-finding and recall. Patients progress into moderate AD when the severity of these symptoms increase. In addition to a significant decrease in shor t term memory, the language difficulties seen in early AD worsen. Moderate AD patients often times create spatially disordered writing and rely on simpler grammatical sentences to express themselves (Forbes et al., 2004). Long term memory loss also develops, and moderate AD patients are prone to wandering and hallucinations as well. The progressive and debilitating decline in cognition eventually leads to the nearly absolute loss of intellect in the advanced stage of AD preceding death. The advanced stage of AD also sees the initial deterioration of motor and sensory skills. Patients in this stage live in a vegetative, bedridden state, unable to care for themselves in any function. Pathological Characterization The prominent lesions characteristic of AD brain pathology, ne uritic plaques and neurofibrillary tangles, were first described by Alois Al zheimer in 1906 but the exact


3 chemical compositions of these lesions were only discovered in the past four decades. The extracellular neuritic pl aques, primarily found in the cerebral cortex and hippocampus, are composed mainly of amyloid peptide (A ), a protein that accumulates in various degrees during normal aging. Several genetic mutations affect the onset of AD by increasing the rate of A production, through mechanisms that will be discussed later. Neurofibrillary tangles cons ist of intracellular paired helical filaments that are left behind as tombst ones when neurons degenerate. Aberrant processing of the amyloid precursor protein (APP) leads to the two forms of A found in neuritic plaques that differ in their amino acid lengths. A 42 the principal component of the core in neuritic plaques, is more susceptible to aggregation than the slightly smaller form of A 40 (Selkoe, 2001). Recognized as a potential precursor to neuritic plaques, diffuse plaques are composed solely of A 42 and generally lack any signs of dystrophic neurons. Th ese diffuse plaques ar e found in significant numbers in the typical areas associated with AD but also appear in brain areas that will never develop mature neuritic plaques. Diffuse plaques are also found in the brains of healthy patients that never show any signs of dementia. Aggregation of A 40 with A 42 leads to the more compact, fibrillar neuritic plaques associated with AD. The accumulation of fibril A in the extracelluar space activates microglia within and surrounding neuritic plaques, which attempt to clear A by phagocytosis and release free radicals and the cytokines IL-1 and TNFSchubert et al., 2000). Astrocytes are found in a concentric ring around the cores of mature plaques, and are activated by the proinflammatory cytokines releas ed by microglia. Astrocytes e xpedite the rate of amyloid


4 deposition and plaque formation by re leasing the inflammatory proteins 1 antichymotrypsin (ACT) and apolipoprotein (APOE) (Potter et al., 2001). Dystrophic neurites are also found amid mature amyl oid deposits, likely as a result of the free radicals and inflammatory mediators rele ased by microglia and astrocytes. This neuroinflammatory cascade that results from the accumulation of A forms the central tenant of the amyloid hypothesis as the causative agent for the cogniti ve decline seen in AD. Neurofibrillary tangles (NFT) accrue when the microtubule-associated protein tau is phosphorylated by intracellular kinase(s). The hyperphosphorylation of tau causes it to uncouple from microtubules and form intracellu lar paired helical filaments (PHF) that constitute the molecular makeup of NFTs. Th e subsequent formation of neurofibrillary tangles from PHFs disrupts normal cell tra fficking and may lead to neuronal death. Neurofibrillary tangles in AD are typically f ound in limbic and association cortices of the brain, often times in association with matu re amyloid plaques (S elkoe, 2001). Although NFTs are considered a pathological hallmark of AD, they are also associated with several other neurodegenerative diseases and occasionally a late-stage AD patient may exhibit low levels of these tangles upon autopsy. Also associated with the insidious na ture of AD is a significant decline in synapses between neurons, as well as the atrophy of neuronal populations in the cerebral cortex and hippocampus. Axonal transport is crucial for synapse viability and accumulation of amyloid has been shown to inhibit fa st axonal transport in cultured rat neurons (Hiruma et al., 2003), giving a plausible explanation for the significant loss of synapses seen in AD. Exposure of human cortical neurons to amyloid in vitro also


5 disrupts calcium regulation, leadi ng to an increase in intracellu lar calcium stores that may enhance neuronal loss via glutamate-induced excitotoxicity (Mattson et al., 1992). This combined loss of synapses and neurons leads to a progressive loss of neurotransmitter systems that are responsible for both s hort and long term me mory, reflecting the behavioral symptoms seen in AD. Genetics of Alzheimers Disease The two types of Alzheimers disease, sporadic and familial, differ only in the age of onset and share similar pathological and be havioral hallmarks. The sporadic form of AD typically has an age of onset over 65 years and represents the majority (>95%) of confirmed Alzheimers patients. Usually the s poradic, or late-onset, form of AD has an unidentifiable cause aside from the possible pres ence of various risk f actors. Patients with a genetic disposition to develop AD represent the familial form and constitute only a small percentage of all AD cases. These pati ents may be diagnosed with AD as early as their thirties, and no later than their sixtie s, depending on the nature of the mutation. Familial AD, or FAD, is linked to autosomaldominant mutations in three causative genes of interest. Each of these mutations disrupt s the normal processing of APP, resulting in increased production of amyloid species. The APP gene is located on chromosome 21, and was initially linked to AD due to the similar neuropathology between AD and the genetic disorder trisomy 21, where the extra copy of APP leads to its increas ed expression and deposition of A (Hardy and Selkoe, 2002). APP is a transmembrane protei n with a long extracel lular N-terminus and a much shorter cytoplasmic C-terminus. During normal APP processing, the protease secretase cleaves APP in the middle of the A domain generating non-amyloidogenic


6 APP fragments. The less common proteolytic cleavage of APP by and secretases generates the amyloid species found in diffuse and neuritic plaques (Zekanowski et al., 2004). The FAD mutations linked to the APP gene change APP processing so that and secretases are more likely to cleave APP and produce amyloidogenic fragments, thus accelerating th e onset of the disease. Specific mutations in the APP gene were only discovered in the past decade. A double missense mutation in the 670 and 671 amino acids of APP was isolated from several Swedish families that exhibited ea rly onset of AD. This double point mutation directly favors the cleavage of APP by secretase and increases levels of A 40 and A 42 (Mullan et al., 1992). Additional missense mutations in the APP gene have also been identified from families of other nati onalities that increase the likelihood of secretase cleavage. In particular, a mutation at amino acid 717 of APP was found in a London family. This mutation increases the cleavage of APP by secretase and raises the levels of A 42 the amyloid species that is esp ecially prone to aggregate. Mutations in the APP gene are relative ly rare however, and subsequent genetic screening of early onset AD pa tients found that the majority of FAD cases can be linked to missense mutations in a presenilin ge ne, PS1. Rarer FAD cases are seen with mutations in another presenilin PS2. Lo cated on chromosomes 14 and 1 respectively, missense mutations in these genes have been li nked to FAD cases in hundreds of families worldwide. The exact molecular actions of th e presenilin genes have been difficult to ascertain, but it is widely accep ted that they influence the secretase cleavage site on APP. It is therefore not surprising that mutations in the presenilins linked to FAD


7 increase secretase activity, leading to elevated levels of A 42 (Tandon and Fraser, 2002). Patients with presenilin mutations linke d to FAD exhibit a 1.5 -to 3-fold increase in neuritic plaques compared to late-onset AD cases (Selkoe, 2001). In addition to being the most common cause of FAD, mutations in the PS1 gene also lead to the earliest age of onset and most aggressi ve form of the disease. Diagnosis of Alzheimers Disease The importance of diagnosing AD at its earliest stage is critical because of its implications in the efficacy of the limited pharmaceuticals and treatments currently used to treat AD. Diagnosing any form of Alzhei mers disease with complete certainty is nearly impossible however until the patient dies and an au topsy can be performed to identify the pathological hallmarks of AD discussed earlier. There are difficulties in recognizing the initial manifestations of Alzheimers symptoms and differentiating between the cognitive decline associated w ith aging and other diseases that cause dementia. Often times the initial diagnosis of AD is made only after other forms of dementia are eliminated and at this point di agnosis is probable at best. Recent work has failed to yield a test battery that can distinguish definitively between the cognitive decline of normal aging, other forms of dementia, and th e initial clinical mani festations of early Alzheimers disease but many recent advances have increased the accuracy in diagnosing AD in more advanced cases. Clinical diagnosis of AD is dependent on several criteria: gradual onset of dementia in the absence of other potentia l dementia-causing disorders between 40 -90 years of age, impaired daily activities, behavioral alterations, family history, neuroimaging, and several biologi cal markers. Some patients ha ve relatively intact daily


8 functioning with cognitive deficits without full-blown dementia. These people fall into the broader categories of MCI, cognitive impairment no dementia, questionable dementia, isolated memory impairment and minimal AD (Nestor et al., 2004). Dementia is diagnosed after a poor sc ore on the Mini-Mental State Examination (MMSE), a test given during a clinical evalua tion that can reveal a decline in memory. Other cognitive deficits seen in word naming and calculations are also seen after further screening. Tests are conducted to rule out co mmon potential causes of dementia such as a vitamin deficiency, hyperthyroidism, neurosyph ilis, or stroke before a diagnosis of probable AD is made. Additional support for an AD diagnosis may come from a positive family history, especially if AD is presen t in first-degree re latives. DNA diagnostic testing can confirm a familial basis of the disease and is useful for screening the risk of family members who are asymptomatic. Commer cial tests are readily available for the more common causative mutations in the presen ilin genes and for variations in the APOE gene that may increase susceptibility for spor adic AD, yet a clinical test for APP remains undeveloped. The presence of one or two copi es of the ApoE4 allele with accompanying dementia also lends support for a more defini tive AD diagnosis, as th is particular allele confers an increased risk to developing the di sease by reasons that will be discussed later (Gaskell et al., 2004). Genetic confirmation of AD using clinical DNA testing is accurate, yet it is limited in its application by the relatively small number of AD cases that have a familial basis. Therefore, the developmentof biological markers that can contribute to the early diagnosis of sporadic AD is underway but has yet to reach widespread clinical application Currently the most widely us ed markers for AD are amyloidproteins and


9 both tau and hyperphosphorylated tau protein levels collected from the CSF, but there are drawbacks to using either marker. AD patie nts are characterized by low levels of A 42 in both blood plasma and the CSF, but there is significant variability in A 42 levels between individuals. Low levels of A 42 may also be found in other di seases, such as depression. This makes it difficult to set the standard ra nge for clinical AD in the overall population using this marker, yet longitudinal studies on individuals are useful (Sobow et al., 2004). The mechanism responsible for a decrease in A 42 concentrations is still open to debate. A recent study found a correlation between lowered CSF A 42 levels and the decreased brain volumes and enlarged ventricles found in AD patients (Wahlund and Blennow, 2003), indicating a possible dilution eff ect and/or decreased production of A 42 It is more likely that the brain acts as a sink and increased deposition of 42 into plaques reduces the concentration of A 42 remaining to diffuse into the blood. The use of phosphorylated-tau protein (p-t au) in the CSF as a diagnostic marker for AD has proven to be far more accurate than A 42 P-tau is consistently seen in high levels in the CSF of AD patients, and is far more specific than A 42 or total tau levels for differentiating AD from other rele vant diseases that may cause dementia. It is also useful for differentiating between geri atric depression and AD, even when there is significant overlap of clinical symptoms (Buerger et al., 2003). Unfortunately, the sampling technique used to collect CSF is highly inva sive and this limits the application of both 42 and p-tau as CSF biomarkers in some circumstances. When applicable and in conjunction with a thorough neurological evaluation and imag ing techniques, the use of


10 these biological markers is very powerful in increasing the accu racy of AD diagnosis however. Some of the more powerful diagnostic markers for AD involve imaging the medial temporal lobe and mon itoring any alterations in par ticular brain areas. Repeated MRI studies reveal significant atrophy in this region, specifi cally in the volumes of the hippocampus and entorhinal cortex, in early to late stage AD patients (Jack et al., 1997, Xu et al., 2000). MRI has also been found im portant in identifying the early conversion of MCI patients to mild AD patients (D eToledo-Morrell et al., 2004), a potentially instrumental finding in diagnosing early AD pa tients at a period wher e current and future treatments can exert their most dramatic effects. An additional tool in the arsenal of AD diagnostics is PET imaging. PET scans traditionally utilize functional imaging to detect changes in brain metabolism by using a radioactively labeled isotope that reveals areas of glucose metabolism. This type of PET scan unfortunately is only usef ul in detecting changes in th e medial temporal lobe in more advanced AD cases (Ishii et al., 1998), yet this type of imaging is useful in studying physiological aspects of AD. Recently a radio active ligand was developed that binds to A plaques and allows imaging of areas of the brain with large numbers of neuritic plaques (Klunk et al., 2004). This new techni que opens the door for easier evaluation of anti-amyloid therapies and as another tool in increasing the accuracy of AD diagnosis. Currently, radioligands for PET imaging of NF Ts remain to be developed yet research into this area is underw ay (Mathis et al., 2004).


11 Risk Factors for Alzheimers Disease The initial onset and later progression of AD is moderated by the presence or absence of various risk factors. The most profound and unavoidable ri sk factor for AD is aging, yet not all of the elderl y will develop AD. Another impor tant risk factor for both late-onset and familial AD is inheritable, the ApoE4 allele (mentioned earlier as a marker used in genetic screening). The ApoE4 allele has a gene dose affect: with two copies of the allele conferring greater risk than possession of one copy of ApoE4 (Veurink et al., 2003). Inheritance of this allele enhances susceptibility for sporadic AD and also decreases the age of onset of the disease. ApoE is involved in cholesterol transport and cholesterol is linked to amyl oid deposition and deposition. Th erefore, the propensity for AD due to possession of ApoE4 is believed to be as a result in a deficiency in amyloid clearance and subsequent increased deposition of beta-amyloid in plaques (Selkoe, 2001). It is also not surprising that high blood cholesterol (LDL) is also a risk factor for AD. The neuronal cell losses characteristi cally seen in AD patients are typically associated with markers for oxidative stress caused by the bodys inflammatory response to neuritic plaques and NFTs. A diet poor in antioxidants, such as Vitamin E or others found in vegetables and fruit, lowers the bodys ability to respond and protect itself from oxidative stress. Low dietary uptake of antioxidants exac erbates neuronal oxidative damage and increases the risk for developi ng AD (Polidori, 2004). High blood levels of homocysteine are also associated with an in creased risk of AD, but it may be as a result of a folic acid deficiency (Quadri et al., 2004). Unlike the associated risks for AD men tioned thus far, an environmentally enriched lifestyle provides protection agai nst AD and decreases the risk of developing


12 dementia with increasing age (Fratiglioni et al., 2004). Su ch an environment includes exposure to intellectually stim ulating activities, as well as profound social and physical activities that are usually absent from most nursing homes. The cognitive benefit from these activities are believed to share the same pathway and all seem to reduce stress, increase both cognitive reserv e and blood flow to the brai n (Fratiglioni et al., 2004). Treatments for AD Current treatments for AD are based on treating the symptoms rather than the disease itself. The commonly prescribed acet ylcholinesterase inhibitors (donepezil, rivastigmine, and reminyl) have some efficacy in slowing down the inevitable onset of neuropsychiatric symptoms in patients with mild to moderate AD (Holmes et al., 2004). These drugs compensate for the decline in memo ry due to the deficits in the acetylcholine neuronal network in the medial temporal lobe. They increase synaptic concentrations of acetylcholine by blocking the enzyme acetylcholinesterase, which is responsible for breaking down the neurotransmitter acetylchol ine. This increases the efficiency of remaining cholinergic neurons that have yet to be disrupted from AD progression. As AD advances these drugs no longer have an affect in preserving memory however. The acetylcholinesterase inhibitors have a relativ ely short window of effectiveness, typically between 1-2 years. The only other FDA approved drug for treating AD symptoms, Memantine, was only recently approved for trea ting moderate to severe AD patients. Memantine is a nonspecific NMDA receptor antagonist that reduces glutamingeric-caused excitoxicity that may be present in the pathogenesis of AD. Used in conjunction with donepezil, memantine has been shown to make signifi cant improvements in cognitive function and


13 daily activities in even severe AD pati ents (Tariot et al., 2004), but like the acetylcholinesterase inhibitors memantin e has a relatively limited duration of effectiveness. II. Animal Models of AD The complex etiology of Alzheimers di sease requires a practical model that closely resembles the ontogeny of the dis ease and can provide in sight into possible therapeutic preventions. Human testing is ge nerally out of the question due to moral implications in testing novel treatments, so researchers turned to animal models. In particular, the mouse lines PDAPP, APP sw and APP/PS1 have been genetically altered to replicate behavioral and pathological aspects of AD. These mice are manipulated by randomly inserting a wild-type or mutant AD transgene into the genome of a fertilized mouse egg. Utilizing a strong promoter that overexpresses the gene in brain-specific areas, these transgenic mice enable the impact of the gene of interest to be evaluated. The short life span of mice (1 to 2 years), while beneficial for research, requires that the promoter overexpress the transg ene of interest so the same symptoms that may take decades to occur in humans will develop in mice (Seabrook and Rosahl, 1999). Numerous mouse lines have been developed, each with their own assets and drawbacks in replicating the disease in humans.


14 PDAPP Model Pathology. The PDAPP mouse model incorporat es a platelet-derived growth factor (PDGF)promoter in neurons in the brain to drive overexpre ssion of a human APP minigene associated with the London type mutation (APP 717V F ) found in some familial AD cases. When compared to wild type littermates, this genomic alteration results in a well characterized, age-depende nt neuropathology exhibiting extracellular levels of and amyloid deposits similar to that found in AD. Diffuse and mature A deposits are seen in the hippocampus, corpus ca llosum, and cingulate cortex as early as 34 months old in PDAPP mice (D odart et al., 2000). Plaque de nsity increases with age in these regions and by 6-10 months numerous plaques are seen (Games et al., 1995; Irizarry et al., 1997). In aged PDAPP mice, in creasing plaque burden is associated with the neuritic dystrophy, cytoskelet al alterations, synaptic dege neration, and gliosis that are also found in AD patients (Schenk et al., 1997 ; Larson et al., 1999). PDAPP mice do not exhibit any neurofibrillary tangles (Masliah et al., 1996) and do not develop the characteristic neuronal loss associated w ith AD however. These mice do show marked hippocampal atrophy by 3 months but this is possibly due to developmental abnormalities, independent of increased A deposition (Dodart et al., 2000). Behavior. Important correlations have been made between the pathological consequences of progressively overexpressi ng this mutant hAPP gene in PDAPP mice and the behavioral results that ensue. These mice can be beha viorally tested at various time points to link cognitive deficits to di sruptions in specific brain areas due to increasing amyloid plaque burden. Initial testing found PDAPP mice develop an agedependent decline in object rec ognition that the authors linked to the increased levels of


15 A plaques in cortical structur es in mice older than 6 mont hs (Dodart et al., 1999). The same study found deficits in spatial memory in 3, 6, and 9-10 month old PDAPP mice using an 8-arm radial maze. Deficits in sp atial memory at the early 3 month time point suggest an age independent decline because this is well before significant amyloid deposition has occurred. In a later paper, the authors attributed this decline in spatial memory to the overexpression of human APP and subsequent abnormal hippocampal formation during development rather than in creased amyloid deposit ion (Dodart et al., 2000). A different study using a novel water maze ta sk revealed an age-related decline in working spatial memory. These PDAPP mi ce were found cognitively intact at 6-9 months, but were impaired at 13-15 months and progressively worsened by 18-21 months (Chen et al., 2000). These authors found this age-dependent decline did correlate with increased -amyloid plaques and linked it to in terruption of synaptic transmission by amyloid burden. Further support for an age -related decline in spatial memory in the PDAPP mouse line came from a recent study in 2004. This study utilized a PDAPP mouse model that does not exhibit abnormal hippocampal atrophy during development. Using a full 6 week battery that examines sensorimotor skills and cognition, Nilsson et al. (2004) found no differences between 2 month old PDAPP mice and the non-transgenic controls. Another group of PDAPP mice were tested at 16 months of age and th ese animals were impaired in Morris water maze and in overall radi al arm water maze performance (RAWM) (Nilsson et al., 2004). Further st atistical analysis of 15-16 month old PDAPP mice found a significant correlation between deficits in Morris water maze, platform recognition, and RAWM performance with the deposition of A in the hippocampus and cerebral cortex


16 (Leighty et al., 2004). Such co rrelations provide evidence that the PDAPP mouse line develops impaired working memory because of accruing levels of A APP sw and APP23 Models Pathology. The APP sw and APP23 mouse models for AD both incorporate a human APP gene with the Swedish double mutation (K670N/M671L). These mutations enhance cleavage of APP by secretase which favors A 42 production. These APP sw and APP23 models differ however because each one utilizes a different promoter that overexpresses APP by varying degrees. This di fference leads to important distinctions between the two models in both pathology and behavior. Initially described by Hsaio et al. (1996), the APP sw mouse line uses a hamster prion protein promoter limited to neurons in that brain that drives fi ve to sixfold higher expression of a human APP695 insert when compared to expression of the endogenous mouse APP gene. The consequences of overe xpressing this insert leads to abnormal changes in the pathology of these mice. Given the nature of the mutations in the insert, it is not surprising that APP sw mice show increasing levels of total A in the brain. Specifically, Hsiao et al. (1996) f ound that the concentration of A 40 was 5-times higher and the concentration of A 42 was 14-times higher when comparing APP sw mice between 2-8 months and 11-13 months of ag e. Rare plaques of insoluble A begin appearing in small quantities as early as 6-7 months in the hippocampus and entorhinal cortex, with diffuse and compact plaques increasing to sign ificant numbers by 12 months of age in the hippocampus and cortex of the brain (Hsiao et al., 1996; Kawara bayashi et al., 2001) eventually resembling similar levels found in AD patients.


17 Importantly, amyloid plaques in the CA 1 area of the hippocampus of 16-month old APP sw mice were found in association with g liosis and neuritic dystrophy (Irizarry et al., 1997). Other studies have also found activ ated microglia in response to amyloid plaques in aged APP sw mice (Frautschy et al., 1998; Be nzing et al., 1999). Another study found an accumulation of prostaglandin E 2 (PGE 2 ) and tumor necrosis factor (TNF), both mediators of inflammati on from gliosis, in APP sw mice older than 10 months of age (Quadros et al., 2003). A further study using 21-25 month old APP sw identified areas of oxidative damage in association with amyloid plaques, likely because of the release of free radicals during the inflammatory proce ss (Pappolla et al., 1998). The presence of these inflammatory proteins and A -induced oxidative damage makes this mouse line an attractive model for testing therapeutics that target the inflammatory cascade seen in AD. Although the APP sw mouse line replicates seve ral aspects of AD, namely amyloidosis, vascular angiopathy, free radical formation, and inflammatory mediators in association with amyloid plaques, it is an incomplete model of AD because these mice fail to develop neurofibrillary tangles and no gl obal neuronal or synaptic losses have been reported. Even in aged APP sw mice, where extensive plaque formation is associated with inflammatory mediators and pro-oxidant activ ity, neuronal populations remain relatively consistent with neuronal number s in non-transgenic controls in areas of the brain such as the hippocampus, where large numbers of ne urons atrophy during the progression of AD in humans (Irizarry et al., 1997). With no clear abnormalities in the brain structures associated with learning and memory in these mice, some researchers have suggested that the increasing soluble levels of A are to blame for the increasing behavioral impairments that APP sw mice exhibit as they grow older. In fact, one study found severe


18 impairment in in vitro and in vivo long-term potentiation (LTP ) in neurons of APP sw mouse hippocampus that correlate with cognitive impairment and rising levels of both soluble and insoluble A (Chapman et al., 1999). The other mouse model that incorpor ates the Swedish double mutation, the APP23 transgenic mouse, employs a murine Thy-1 promoter element to overexpress a mutant APP751 gene in neurons in the br ain. These mice have a seven fold higher expression of the mutant APP insert th an normal mouse APP (Dodart et al., 2002). APP23 mice develop A plaques in the cortex at 6 months, much sooner than APP sw mice (Sturchler-Pierr at et al., 1997). A plaque numbers increase with age and spread extensively in this model to the neocor tex, hippocampus, white matter, and thalamus. Associated with these pl aques are dystrophic neurons activated microglial and astrocytes. These mice also develop congoph ilic angiopathy, where large deposits of A 40 are found in conjunction with the brai n vasculature (Calhoun et al., 1999), a trait found in some 90% of AD patients (Vinters, 19 87). In striking dissimilarity to the APP sw mouse model, APP23 mice e xhibit a 14% loss of CA1 pyr amidal neurons in the hippocampus in 14-18 month old mice, alt hough no losses are seen in the neocortex (Calhoun et al., 1998). The authors attribute this loss directly to the formation of dense A plaques. This model also fails to develop neurofibrillary tangles. APP sw Behavior. An important component of the behavioral characterization of transgenic mice is an assessment of any sensor y or motor deficits that may influence the performance of the mice in cognitivebased tasks. Accordingly, the APP sw mouse failed to show any deficits in a visi ble platform tasks at either 6 or 9 months of age, indicating that visual acuity is not impaired in these mice (Hol comb et al., 1999). Further


19 sensorimotor evaluation, assessed through the use of various agility tasks and a Preyer reflex test, also failed to detect any si gnificant disturbances (Holcomb et al., 1999; Holcomb et al., 1998). In a later study, APP sw mice at 3, 9, 14, and 19 months of age were subjected to a full battery of sensor imotor tasks. The authors concluded that younger APP sw mice are not impaired in motor f unction overall, and older transgenic mice show impairment similar to non-transg enic controls (King et al., 2002). Recently, Arendash et al., (2004) repor ted that 6 month old APP sw are impaired in a balance beam task, but perform normally in the elevated maze task for anxiety and other sensorimotor tasks, such as string agility. Additionall y, Leighty et al. (2004) found no correlation between the balance beam task, a measure of sensorimotor skills, and cognitive performance. The lack of any sensory or motor abnormalities in APP sw mice, or significant correlation with cognitive perfor mance, validates linking any performance issues in cognitive-based tasks to disruptions in memory, not to the physical attributes of the mice. The Y-maze alternation task has elucidated deficits in mnemonic processing in the APP sw mouse, but incons istently so. Hsiao et al. ( 1996) first found 3 month old APP sw mice performed similar to controls, yet 9 month old Tg + mice showed significant impairment in spontaneous alternation in this task. King et al. (2002) found APP sw mice aged 3, 9, 14, and 19 months had an overall reduced alternation in the Y-maze task when all age groups were analyzed collectivel y. In a later study, Are ndash et al. (2004) found APP sw mice were impaired overall in sponta neous alternations at 5 and 8.5 month time points collectively. On the other hand, Holcomb et al. (1999) reported APP sw mice were impaired in alternation behavior at 3 months yet were unimpaired at 9 months of


20 age. This task is likely not as sensitive as some of the other cognitive-based tasks because Y-maze impairment remains inconsistent in this mouse model, although differences in methodologies and genetic backgrounds may pl ay a role in these discrepancies. Not surprisingly, 15-17 month old APP sw mice also showed significant T-maze alternation impairment (Chapman et al., 1999) The authors correlated the poorer T-maze impairment to their in vivo findings of impaired long-term potentiation in both the CA1 and dentate gyrus regions of the hippocampus from the same mice and suggest that rising A levels in these aged mice acts directly to infer synaptic deficiencies evident in reduced LTP. Unfortunately at this late age, the authors could not discriminate between the effects of soluble or insoluble A although the authors hint diffusible forms of A are most likely responsible. Th is group also showed the APP sw mouse model lacks deficits in sensorimotor attributes, but onl y used an open field task to claim this. Initial cognitive testing of APP sw mice using the Morris water maze revealed no differences in learning and memory in spatia l reference tasks between non-transgenic and transgenic mice at 3 months of age (Hsiao et al., 1996). At a later age in this study however, 9 month old APP sw were found impaired in their escape latency in the same task and also spent less time in the platforms quadrant in the probe trial, indicating these mice did not learn the location of the platfo rm. Recently in 2004, however, Arendash et al. found impaired acquisition and retention performance in the Morris water maze as early as 5.5 months of age, suggesti ng the initial app earance of small A oligomers are responsible for impairment of both refere nce learning and memory. A previous study done by King and Arendash (2002) reported their mice were unimpaired in Morris water maze even up to 19 months of age. These contrasting results from the same lab were


21 suggested to be caused by the different b ackground strains of the mice used in these studies or by changes in their genetic backgrounds caused by multiple generations of inbreeding. These differences enabled the e ffects of the transgene to become more evident in the later study. Further evidence for the role of cognitive disturbances by small A oligomers came from a study that reported impaired reference memory in the Morris water maze task at 6 months of age in APP sw mice (Westerman et al., 2002). Although one study using the Morris water maze failed to replicate cognitive impairment in 6 or 9 month old APP sw mice, this is likely because of differences in testing paradigms and the genetic backgrounds that exist between mouse co lonies in different la bs (Holcomb et al., 1999). In summary, the consensus of the behavior al studies thus far indicate Morris water maze impairment in APP sw mice begins around 6 months of age. Using a circular platform task adopted from rat behavioral studies, King et al. (2002) found 3, 9, 14, and 19 month old APP sw mice perform comparably to nontransgenic age-matched controls, with both groups showing a learning effect that reduced the number of errors made over the 7 day te sting period. Arendash et al. (2004) reported no differences between 6 month old APP sw mice and non-transgenics in circular platform performance as well. Testing by Pompl et al. (1999) revealed impairment in 7 month old APP sw mice during a reversal learning phase of the circular platform task however. APP sw mice learned the location of the escape hole in relation to various cues as well as nontransgenic mice; however, when the escape hole was moved, tran sgenic mice showed increased errors and escape latency that failed to improve. This is a different variation of the task however, and the literature is consis tent in reporting a lack of impairment of APP sw mice in the standard circular platform task at any age.


22 Additionally, APP sw mice repeatedly show poor performance in other cognitive tasks. Hsiao et al. (1996) reported impaired platform recognition latency in 9-10 month old APP sw mice that coincided with increases in soluble A King et al. (2002) also showed APP sw mice at 9 months of age were signif icantly slower in locating a visible platform, and both of theses studies perf ormed the platform recognition task after conducting the Morris water maze task. Recently a similar behavior paradigm using the platform recognition task identified imp airment even earlier in 6 month old APP sw mice (Arendash et al., 2004). In c ontrast, Westerman et al. ( 2002) found unimpaired platform recognition performance throughout the 2 year lif e span of the mice, but this test was performed before Morris water maze te sting. Impaired escape latency for APP sw mice in the platform recognition task thus has been suggested to reflect difficulties in switching between the spatial (cued) strategy used in Morris water maze to a search/recognition strategy that ignores the prev iously learned extra-maze spa tial cues. This explains why APP sw mice are impaired in the platform recogn ition task after they have already learned the platform location in the Morris water m aze task, but have not been impaired in the platform recognition task if it is done prior (Arendash et al., 2004, Westerman et al., 2002). The radial arm water maze (RAWM) is a sensitive working-memory task used to evaluate both short term memory and delaye d memory recall. The first testing of APP sw mice in the RAWM task found 15.5 month old mi ce made significantly more errors in the trial 5 component of the task when compared to non-transgenic controls, indicating these mice were deficient in working memory (Morgan et al., 2000). RAWM testing of younger 6.5-7 month old APP sw mice revealed that these animals made more errors


23 overall in both the trial 4 and trial 5 com ponents of the task, further highlighting the deficits these mice have in working memory (Arendash et al., 2004). The consistent working memory impairment these mice show in this task by 6.5 months of age indicates the oligomeric form of A is likely responsible for thei r cognitive impairment, as these young mice have yet to develop amyloid plaques. In summary, APP sw mice do not develop any gross sensorimotor deficits that would otherwise impair performance in cogni tive tasks. These mice consistently develop significant impairment in reference learning/mem ory that is apparent in the Morris water maze as early as 5.5-6.5 months and is likely caused by the increasing amounts of soluble A oligomers. The difficulties in switching es cape strategies are also apparent for 9 month or older APP sw mice in the platform recognition ta sk, which is reflected in their increased escape latencies. APP sw mice also exhibit working memory impairment in the RAWM task at 6.5 months of age, as these mice make more errors in the trial 4 and 5 components of the task. Based on these overall behavioral findings, it is reasonable to expect 8-9 month old APP sw mice will have impaired reference learning/memory, impaired recognition/identification, and impair ed working memory evident in the tasks discussed thus far. The APP23 mouse model has undergone be havioral characterization as well. A behavioral characterization of the APP23 model was done by Kelly et al. (2003) at 3, 18, and 25 months of age using passive avoidan ce tasks, platform recognition, and small and large Morris water maze tasks. Age-related impairments in passive avoidance and in small pool acquisition were reported, while APP23 mice were found to be impaired at every time point in the larger Morris wate r maze during acquisition testing. The authors


24 also reported visual deficiencies were not re sponsible for this impaired spatial learning by using a platform recognition task in the larg er pool. An age-related decline in Morris water maze and probe trial performance was reported also at the early age of 3 and 6 months, while 2 month old animals were unimp aired when compared to non-transgenic controls (Van Dam et al., 2003). Further studies have also reported pr ogressive deficits in acquisition (learning) of the Morris water m aze task in both 16 month (Lalonde et al., 2002) and 2 year old APP23 mice (Dumont et al., 2004). In addition, increased exploratory behavior was dete cted in 2 year old APP23 mice through the use of the open field task, while anxiety was decreased in the open arm (+) maze task (Dumont et al., 2004). The repetitive results of impaired spat ial memory in APP23 mice in the Morris water maze indicates these animals represent an effective model for behavioral testing in AD, yet the behavior of these mice in working memory tasks, such as the RAWM, that examine long term memory remains to be determined. PSAPP Model Pathology. The PSAPP, or APP sw /PS-1, mouse model is a combination of two mutations associated with familial Alzheimers disease. The same Swedish mutation used in APP sw mice that enhances -secretase activity on APP and a PS-1 mutation that favors the enzymatic cleavage of APP by -secretase are both overexpressed in this model resulting in an exponential increase in A levels with aging, even when compared to that of transgenic APP sw mice (Holcomb et al., 1998). Studies have shown that PSAPP mice begin showing A deposits in both the hippocampus and cortex between 3-6 months of age (Takeuchi et al., 2000), with significant numbers present by 9-12 months of age (Borchelt et al., 1997; Holcom b et al., 1998). Although deposits of A resemble


25 those found in AD, compact plaques of fibrillar A have been reported to form before diffuse plaques in the PSAPP model, wher eas this trend is opposite in human AD (Gordon et al., 2002). Nevertheless, by 12 mont hs of age the compact plaques in PSAPP mice are surrounded by dystrophic neurites, reactive microglia and GFAP-expressing astrocytes, indicating that a neuroinflammatory response is present (Gordon et al., 2002). In depth analysis of the inflammatory response in PSAPP mice found that amyloid plaques increase continuously with age up the latest age studied, 25 months (Matsuoka et al., 2001). These authors found cyclooxygenase-2 (an inflammatory response protein) in conjunction with the astrocyt es surrounding amyloid plaques. Complement component 1q was identified with the microglia associat ed with amyloid plaques, which indicates microglia may be attempting to clear amyloid via the complement pathway. Like the other AD animal models discu ssed thus far, the PSAPP mouse line also fails to develop neurofibrillary tangles. Take uchi et al. (2000) repor ted a non-significant loss in cortical and CA1 neurons of the hippocampus in 12 month old PSAPP mice, yet this study used a low sampling number. A fu rther study of 22 mont h old PSAPP found a 38.5% reduction in CA1 pyramidal neurons (Sad owski et al., 2004). Deficits in LTP have also been found as early as 3 months in PSAPP mice, when amyloid plaques and deficits in short term memory first a ppear (Trinchese et al., 2004). PSAPP Behavior. The first study to examine behavi oral aspects of PSAPP mice found that by 3-4 months these mice had signi ficantly impaired alternation performance in the Y-maze task, months before the initial deposition of A has been detected (Holcomb et al., 1998). The same lab found th at this deficit in Y-maze performance persisted with increasing age, as PSAPP mice we re also impaired at 6 and 9 months of


26 age (Holcomb et al., 1999). This same study also failed to detect any deficits in spatial working memory at 6 or 9 months of age using the Morris water maze. These results show that behavioral deficits are lik ely due to the presence of soluble Using a full behavioral battery, a later study found that PSAPP mice showed no change in spontaneous Y-maze alternations at either 57 or 15-17 months of age (Arendash et al., 2001). This same study also showed normal Morris water maze performance at 5-7 months of age but an aged-related impairm ent in this task was found later at 15-17 months. Another study investigating the eff ects of a lifelong imm unization with human A found unvaccinated PSAPP mice were impaired in both Morris water maze and the radial arm water maze (RAWM) at 4-6 months and 15-16 months of age (Jensen et al., 2005). Similar impairments in RAWM at 15.5 m onths were reported in another study of PSAPP mice (Morgan et al., 2000). The findings of cognitive impairment at such an early age in this same colony of mice was attributed by the authors to multiple generations of crossbreeding that enhances the susceptibility of the mice to the effects of mutant APP overexpression. Early PSAPP working-memory deficits at 3-4 months of age in the RAWM and impaired reference-memory in the Morris wa ter maze at 6-8 months were also reported by Trinchese et al. (2004). The working-memo ry impairment was evident in the number of errors the PSAPP mice made in the RAWM that strongly correlated with A levels and amyloid burden in these mice. A furt her correlation was made between RAWM errors and synaptic deficiencies in LTP at 6-8 months of age. The authors propose this lends additional support to the idea that increasing A levels and amyloid burden impair LTP, thus reducing the animals perfor mance in tasks based on cognition.


27 Limits of Animal Models for Alzheimers disease In addition to the various pathological aspects of AD that fail to materialize in AD transgenic mice (such as neurofibrillary tangles or global neuronal loss that were mentioned earlier for each specific model) are differences in mouse and human pathology that possibly may skew the generalizations ma de after a potential therapeutic is tested. Specifically, A plaques in both APP23 (Kuo et al., 2001) and APP sw mice (Kalback et al., 2002) were found to be soluble in SDS-containing buffers whereas human plaques are not, indicating mouse plaques are le ss dense. This distinction is important because it is believed to be responsible for the weaker complement response to plaques and the resultant weaker inflammato ry cascade found in APP23 mice (Schwab et al., 2004). Knowing that human fibrillar A plaques are neurotoxic from in vitro studies (Lorenzo et al., 2000), a less compact form of these plaques in the mice may also explain the lack of neuronal losses in animal models of AD. Transgenic mice are also incomplete in th eir behavioral replicat ion of the disease. Many of the behavioral aspects of early Alzhei mers disease are difficult to ascertain in humans even through the use of cognitive test s such as the MMSE that examine various aspects of memory. Semantic memory, such as word or event recall, is usually the first apparition signaling Alzheimers disease in humans and cannot be assessed in a mouse animal model. Obviously the language difficu lties which surface in moderate AD are also problematic to examine in mice. Mouse anim al models are thus limited to behavioral measures that examine spatial learning/memo ry, recognition/identific ation, exploratory behavior, sensorimotor skills, and anxiety, wh ich may only give some insight into the onset of depression, motor skills, or memory impairment of this disease in mice.


28 Given these limits in replicating all asp ects of Alzheimers disease, transgenic mice still have been paramount in our understand ing of the disease a nd the possibility of developing therapeutics that la y the foundation for future clin ical testing looks promising through the use of these mice. III. Caffeine Consumption and Alzheimers Disease There is little doubt that the widespre ad consumption of caffeine by the global population makes it an important consideration when evaluating dietary influences on the etiology of human diseases. Caffeine can be found in popular foods and beverages such as tea, coffee, cocoa, chocolate, and soft drinks that constitute dietary items consumed sometimes as a chronic staple of the Western diet. The average caffeine consumption in the US is equivalent to 1-2 cups of coff ee a day, or 168 mg/person/day (Fredholm et al., 1999). Caffeine may also be found in certain aspirins and other overthe-counter drugs, although these would not represent a sour ce of long-term caffeine ingestion. Caffeines stimulatory effects on behavi or and attention have been known for some time, but its effects on other aspect s of human neurobiology have only recently begun to be explored. Epidemiological stud ies indicate that sufficient daily caffeine intake throughout life may be neuroprotectiv e in both Parkinsons disease (Ross et al., 2000) and Alzheimers disease (Maia a nd Mendonca, 2002). Notably, Maia and Mendonca found that AD patients consumed ma rkedly less caffeine (74 mg/day) during the twenty years preceding diagnosis of AD when compared to age-matched controls


29 (199 mg/day). Daily caffeine intake include d any potential sources of caffeine and was not solely restricted to caffe ine ingested from coffee. This information suggests that chronic caffeine intake during the middle ye ars of life may delay or even prevent altogether the onset of the behavioral symptoms seen in sporadic Alzheimers disease, an important finding. The molecular mechanism(s) that grants caffeines neuroprotective effects remains to be clearly elucidated, but many po ssibilities have been suggested. Once this mechanism has been identified, caffeine-base d therapeutics could become useful in treating neurodegenerative diseases because of their availability and general lack of longterm side effects. Pharmacological Profile of Caffeine Whether taken orally or administered intravenously, the pharmacokinetics of caffeine are identical in humans a nd animals (Arnaud, 1993). Likewise, the gastrointestinal absorption and bioavailabil ity of caffeine reaches 99-100% within 45 minutes of ingestion in both human and animal models (Bonati et al., 1984, Blanchard and Sawers, 1983a). Peak plasma level of caffeine is reached between 15 and 120 minutes after ingestion in humans, with a half-life that ranges from 2.5 to 4.5 hours in humans (Arnaud, 1987) and only 0.7 to 1.2 hours in rats and mice (Bonati et al., 1984). It is also known that the plasma half-life of caffeine remains relatively unchanged in both young adults and the elderly (Blanchard and Sa wers, 1983b). On average, a typical cup of coffee provides a caffeine dose of 0.4 to 2.5 mg/kg which gives a peak plasma concentration of caffeine of 1 to 10 M (Fredholm et al., 1999). This information, once accounting for weight and metabolic differences between animal models and humans,


30 allows for interpolations for dose-dependent effects between caffeine studies in animal models and humans. Caffeines hydrophobic nature allows for rapid absorption through all biological membranes, hence its rapid absorption into the bloodstream. This same tendency also allows caffeine to freely pass through the blood brain barrier (Tanaka et al., 1984), an obstacle that limits the design and size of ma ny other potential neurotherapeutics. Once absorbed into the blood stream, caffeine is al so broken into its main derivatives by the liver. Caffeines major metabolites in rodent s and humans with bioactivity include 1,3dimethylxanthine (theophylline) and 1,7-dimet hylxanthine (paraxanthine). Both of these compounds were found to mimic some of caffein es effects in the CNS and which will be discussed in greater detail la ter on (Benowitz et al., 1995). Th eophylline in particular is of interest because of its common use as a treatment for asthma and other diseases that increase bronchial constriction. Molecular Actions of Caffeine Caffeine has been used experimentally for some time to study its inhibitory effect on cyclic nucleotide phosphodiesterase isozym es of the brain and other tissues of the body (Vernikos-Danellis and Harris, 1968). The ca ffeine dose needed in order to achieve this effect, however, is within the millim olar range which far exceeds typical human plasma levels of caffeine after ingestion. This same range is also required to release intracellular calcium stores via activation of ryanodine receptors (McPherson et al., 1991). In addition, caffeine plasma levels in excess of 500 M, or more than 50 cups of coffee, are toxic, likely because of caffeines blockade of GABA A receptors at this extreme range (Fredholm et al., 1999). Therefore, it is reasonable to assume that neither


31 of these actions of caffeine can be attributed to its neuroprotective effects that are being investigated. The only other molecular action of caffeine that can be achieved within physiological doses in the CNS is adenosine receptor antagonism, a mechanism that is shared by both theophylline and paraxanthine. In fact, theophyllines affinity for adenosine receptor antagonism is three to five times higher than caffeine (Fredholm et al., 1999). Theophylline has also been shown to enhance hippocampal LTP, yet the dose needed for this effect is in the 100 to 1000 M range which is beyond that of typical caffeine consumption (Tanaka et al., 1990). Adenosine. As the main constituent of ATP, adenosine is present both intracellularly and extracellularly th roughout the body. Although adenosine is an important modulator of neurotransmissi on, it is not considered a classical neurotransmitter because it is neither stored in vesicles nor dependent on Ca+ for release (Fisone et al., 2004). Instead, adenosine is transported between the cytoplasm and extracellular space through equilibrative nucleoside transporters. The direction of this transport is dependent on the adenosine c oncentration gradient in both sides of the membrane (Gu et al., 1995), which under norma l conditions facilitate s the intra-cellular transport of extracellular adenosine as adenosine is incorporated into AMP intracellularly via the enzyme adenosine kinase. During times of physiological stress, such as ischemia or hypoxia, intracellular leve ls of adenosine rise due to hydrolysis of ATP which subsequently fuels an increase in the extracellular concentration of adenosine (WallmanJohansson and Fredholm, 1994). Aside from adenosine efflux from the cytoplasm, a second minor source of adenosine exists in the extracellular space that involves the


32 conversion of released adenine nucleotides into adenosine by several enzymes (Zimmerman and Braun, 1999). Rising extracellular levels of adenosine are thus viewed as a protective response from metabolic injury and various therap eutics have been proposed recently to specifically raise extracellular adenosine leve ls. In particular, the drug propentofylline has been forwarded as a potential treatmen t for Alzheimers disease, and it showed promising results in a clinical study (Kittner et al., 1997). One role of this drug is blockade of the equilibrative adenosine transporters, which leads to an increase in extracellular adenosine levels because of the conversion of adenine nucleotides into adenosine mentioned earlier. The eleva tion of extracellular adenosine levels downregulates activated glial cells and leads to the restoration of altered calcium homeostasis (Ringheim, 2000). Likewise, the dr ug dipyridamole blocks adenosine uptake and inhibits the enzyme adenosine deaminase th at breaks down adenosine, thus leading to increased extracellular concentrations a nd has been shown to block the enhanced vasoconstriction induced by soluble A by increasing cGMP activity (Paris et al., 1999). Adenosine itself is also admi nistered commonly by paramedics as a treatment to slow atrial tachycardia by increas ing conduction time through the AV node. Finally, it has also been found that long-term ingestion of caffein e in rats leads to significantly increased levels of adenosine in blood plasma in a dose-dependent manner through an unknown mechanism (Conlay et al., 1997). Although ad enosine does not cross the blood brain barrier, it can be assumed a ri se in brain adenosine levels should accompany rises in plasma adenosine levels because caffeine can enter the brain from plasma and exert similar effects on adenosine receptors there as well.


33 Adenosine Receptors. Currently, four G-proteincoupled receptor (GPCR) subtypes have been identified to have ad enosine affinity. These subtypes, named A 1 A 2A A 2B and A 3 are expressed throughout the brain in particular and in both the central and peripheral nervous system (Fredholm et al., 1994) ; they are colle ctively referred to as the P1 (adenosine selective) receptors. The high affinity receptors A 1 and A 2A are typically of interest because they are most likely res ponsible for the neuromodulatory effects of adenosine at physiological levels while the lower affinity A 2B receptors have been relatively ignored because they are most likel y only activated when adenosine levels are increased (Fredholm et al., 1999; Fisone et al., 2004). The low affinity A 3 receptors are sparsely distributed in humans and are little affected by caffeine or its metabolites in rats (Fredholm et al., 1999). Although A 2B and A 3 receptors cannot be completely excluded from having a neuroprotective effect in rela tion to caffeine, this is simply because evidence that might support such a role is lacking so these receptors are not included in this thesis. Refer to Figure 1 at the end of this section for an overview of caffeinemediated effects by adenosine receptor antagonism. The A 1 type adenosine receptors have high affinity for both caffeine and adenosine. These receptors are widely distribut ed in the brain, with higher expression in the cerebral cortex, cerebellum, hippocampus, an d in the dorsal horn of the spinal cord (Ribeiro et al., 2003), while lower expression levels of this receptor are also found in the basal ganglia (Rivkees et al., 1995). These receptors (along with the A 3 receptors) activate inhibitory G-proteins (both G i and G o ) which may lead to the inhibition of adenylyl cyclase, closure of Ca 2+ channels (MacDonald et al., 1986), and activation of K+ channels (Trussel and Jackson, 1985). Inhibition of adenylyl cyclase ultimately leads


34 to decreased production of cAMP thus activation of these re ceptors is generally thought to provide an inhibitory effect on sec ondary messenger systems while stabilizing the cellular membrane by blocking Ca 2+ influx. The A 1 receptor is present on the membranes of neurons, microglia, and astrocytes (Ongini and Sc hubert, 1998). The majority of neuronal adenosine A 1 receptors are located on presynaptic nerve termin als where they provide inhibition of neurotransmitter release after activation by adenosine (Fisone et al., 2004), although some post synaptic A 1 receptors may also exert an inhibitory effect (Ribeiro et al., 2003). Most importantly, activation of A 1 receptors has been shown to limit neuronal release of the excitatory amino acid glutamate (Fla gmeyer et al., 1997) and decreases in acetylcholine release have also been repor ted (Brown et al., 1990). Activation of A 1 receptors in the striatum was also linked to inhibition of the D 1 receptor-mediated increase in adenylyl-cyclase activation (A bbracchio et al., 1987), giving an explanation for the reports of decreases of striatal extracellular dopamine levels following A 1 receptor activation (Okada et al., 1996). Caffeine administration woul d disinhibit these neurons, leading to an increase in neurotransmitter release (see Figure 1). The inhibitory effect of adenosine A 1 receptor activation on glutamate release and Ca 2+ influx likely is responsible for the neuroprotection (particularly in hippocampal regions) seen when animal models of cerebral ischemia are administered A 1 selective adenosine receptor agonist s (Rudolphi et al., 1989). The stabilization of Ca 2+ homeostasis after A 1 receptor activation raises the depolariz ation threshold needed to remove the Mg 2+ blockade in NMDA receptors (present in la rge numbers in hippocampal neurons) leading to a reduced risk of excitotoxicity (Ongi ni and Schubert, 1998). Treatment with an A 1


35 selective adenosine receptor antagonist on th e other hand, is shown to exacerbate damage in the hippocampus during ischemia because of the increased exposure to excitotoxicity (Rudolphi et al., 1997). The A 2A type adenosine receptors also have high affinity for caffeine and are activated by the nanomolar concentrations of adenosine present during normal physiological conditions in the brain (Fisone et al., 2004). These receptors are highly expressed in the basal ganglia, particularly in the striatum and the olfactory bulb, while much lower levels of this receptor are found in the hippocampus and cortex (Sebastiao and Ribeiro, 1996). The A 2A receptors activate stimulatory G-proteins (G s in the periphery and G olf in the striatum) which in turn activ ate adenylyl cyclase (Herve et al., 2001) and may also activate voltage-sensitive Ca 2+ -channels in certain cells (Fredholm et al., 1999). It has been shown in some hippocampal neurons that A 2A receptor activation upregulates Ca 2+ uptake via class A calcium channels, through the activation of adenylyl cyclase signaling (Goncalves et al., 1997). Notably, A 2A adenosine receptors sometimes are located with A 1 receptors on the same cells, indicat ing these different receptor types may have opposing roles when on the same cell. Indeed, in the striatum (a brain area where A 2A receptors are high and A 1 receptors are low) activation of the A 2A receptor leads to a desensitization of the A 1 receptor (Dixon et al., 1997). The end result of adenosine receptor activation thus may depend on the densities of each receptor type in that particular brain area. A 2A receptors are located on both neurons and microglia, but not on astrocytes (Fiebach et al., 1996). A 2A receptors are also present on coronary arterial walls and upon activation cause vasodilation (Linden, 2001). In addition, A 2A receptors are also located


36 in the periphery on bone-marrow derived cells; such as neutrophils, monocytes, macrophages, platelets, and mast cells, where in vitro activation of these receptors has been shown to reduce the rel ease of reactive oxygen species and the pro-inflammatory cytokine TNF(Sullivan, 2003). An in vivo study confirmed that genetic inactivation of A 2A receptors in the periphery results in the loss of a negative feedback mechanism that would otherwise limit a systemic inflammato ry response and reduce tissue damage (Ohta and Sitkovsky, 2001). This suggests that A 2A receptor activation potentially might reduce the pro-inflammatory component of the amyl oid cascade in the brains of Alzheimers afflicted patients. Expression of A 2A receptor mRNA on glial cells in the brain is limited however (Schiffman et al., 1991), thus their role in any A 2A -modulated actions is unlikely to be significant. Theref ore, blockade of glial A 2A receptors by caffeine would also be limited so it should not have a noticeable eff ect on inflammation in Alzheimers disease (see Figure 1). Activation of A 2A receptors by selective A 2A receptor agonists has shown significant increases in extracellular glutamate levels in the striatum (Popoli et al., 1995) and cerebral cortex (ORegan et al., 1992), pr oviding evidence that activation of striatal A 2A adenosine receptors gives a detrimental effect during times of ischemia and/or excitotoxicity in dir ect contrast with A 1 receptor-mediated protection. Accordingly, A 2A receptor blockade by selective A 2A adenosine receptor antagonists has been found to reduce neuronal striatal damage induced by the excitotoxin quinolic acid (Popoli et al., 2002) and A 2A receptor KO mice have been reported to be protected against MPTPinduced neuronal damage in the striatum (Che n et al., 2001). With this information in mind, any neuroprotective effects of A 2A antagonists against Alzheimers are likely due


37 to vasodilatation in the peri phery, rather than in the br ain itself because excitotoxic damage is not indicated in st riatal areas in patients with this particular disease. Particularly in Parkinso ns research, neuronal A 2A receptors have received intense interest because of th eir roles in dopaminergic transmissi on in the dopamine rich areas of the striatum. The majority of striatal A 2A expression is found in association with the inhibitory dopamine activated D 2 receptors on the GABAergic medium-sized neurons of the indirect pathway that project from the striatum to the gl obus pallidus and subthalamic nucleus (Ferre et al., 1997). Under normal conditions, dopamine is the major modulator of this pathway, with A 2A receptors playing an opposi ng role to activation of the inhibitory D 2 receptors. When dopamine levels diminish in Parkinsons disease, this GABAergic/globus pallidus/subthalamic nucleus pathway is overactiv e, contributing to the motor abnormalities present in this disease. Thus A 2A specific antagonists have been forwarded to potentially correct this imbalance, in addition to their neuroprotective effect mentioned earlier. While this information is important for potential therapeutics against the degeneration of the dopaminergic system s een in Parkinsons disease, it is unlikely to have much bearing in Alzheimers disease.


Caffeine Intake A1Receptor Antagonism A2AReceptor Antagonism Unknown cAMPGlutamate releaseGABADopamine release Glial ActivationVasoconstriction cAMPGlutamate releaseAcetylcholine releaseCa2+Influx Dopamine release UnknownHippocampus/CortexStriatumGlobal CNSFigure 1. Receptor-mediated effects of caffeine intake Caffeine Intake A1Receptor Antagonism A2AReceptor Antagonism Unknown cAMPGlutamate releaseGABADopamine release Glial ActivationVasoconstriction cAMPGlutamate releaseAcetylcholine releaseCa2+Influx Dopamine release Unknown Caffeine Intake A1Receptor Antagonism A2AReceptor Antagonism Unknown cAMPGlutamate releaseGABADopamine release Glial ActivationVasoconstriction cAMPGlutamate releaseAcetylcholine releaseCa2+Influx Dopamine release UnknownHippocampus/CortexStriatumGlobal CNSFigure 1. Receptor-mediated effects of caffeine intake Immediate Effects of Caffeine Intake Physiological Changes. The initial absorption of caffeine into the bloodstream creates widespread changes throughout the CNS. Acute doses of caffeine lead to increases in cerebral energy metabolism in areas of the brain responsible for motor activity and the sleep cycle, and this response is not abolished by any appearance of tolerance (Fredholm et al., 1999). Surprisingly, this increase in cerebral metabolism is also accompanied by a decrease in cerebral blood flow due to the cerebral vasoconstriction induced by caffeine seen during rest (Higashi et al., 2004). Caffeines effect on alertness is positive however, so it is likely that any increases in metabolism with reduced blood flow are met with increased glucose utilization. Interestingly, 1-5 mg/kg doses of caffeine in rats leads to widespread increases in glucose utilization throughout the brain, including areas of the cortex, hippocampus, and central components of the nigrostriatal dopaminergic system (Nehlig and Boyet, 2000). A caffeine-induced 38


39 increase in metabolism coupled with decrea sed blood flow could potentially lead to increased levels of extr acellular adenosine in an ischemic-like response. In the absence of caffeine, adenos ine levels increase during prolonged wakefulness which suppresses cortical ac tivity by activating the inhibitory A 1 receptors on mesopontine cholinergic neurons (Rainnie et al., 1994). The incr ease in cortical activity following caffeine intake eviden t in the electroencephalogram (EEG) as decreases in delta and theta waves and an incr ease in alpha and beta waves (Patat et al., 2000) is thus linked to caffeines antagonism of A 1 receptors on these particular neurons. Caffeine also decreases GABA release in th e globus pallidus, leading to increased activity in this region as we ll (Fredholm et al., 1999). Furthe rmore, a microdialysis study with rats found that caffeine administra tion dose-dependently raised dopamine and acetylcholine concentrations in the prefrontal cortex (Acquas et al., 2002). The authors also reported the increases in dopamine c oncentrations were lost after tolerance developed to chronic caffeine treatment, while caffeines affects on acetylcholine were not. This information lends support to the reported positive affect of caffeine on alertness that is not lost in chronic caffeine users. Behavioral Effect. Although acute use of lower doses of caffeine is known to promote locomotor effects in caffeine intolera nt rodents, the exact mechanism for this action remains to be clearly pinpointed to one receptor. Acute doses of caffeine given in mice in particular are reported to decrease ex ploratory behavior in the open field (Meyer and Caston, 2004) and increase anxiety in th e plus-maze task (Silva and Frussa-Filho, 2000). Given that A 2A receptors are found in particular abundance in the striatum (an area of the brain profoundly involved in locomoti on) it has generally been assumed that


40 blockade of A 2A receptors following caffeine administration is mainly responsible for the changes in locomotor activity (see Table 1). Concordantly, it is not surprising that A 2A receptor KO mice are resistant to the stimulat ory effects of caffeine (Halldner et al., 2004). Caffeine also is reported both anecdot ally and experimentally to provide an immediate positive effect on learning and memor y. It is interesting to note that a study found an acute dose of caffeine administered to caffeine-nave rats after training in the Morris water maze improved their reference memo ry in the probe trial at doses of 0.3-10 mg/kg, whereas a higher does of 30 mg/kg ha d no effect on performance (Angelucci et al., 2002). Acute caffeine doses of 0.1-0.3 mg/kg also were also found to reverse memory disruption in the two-way activ e avoidance task induced by MPTP injections in rats (Gevaerd et al., 2001). Additional studies found that administration of an A 1 specific receptor antagonist MDL102503 to rats re verses scopolamine-induced memory impairment in the water maze. In addition, another A 1 specific receptor antagonist (KFM19) also improved performance in other cognitive based tasks, such as the Y-maze (Jacobson et al., 1996). Castellano (1976) also found a caffeine dose of 1 mg/kg to mice improved the learning and consolidation processes in a Y water maze task. Although these studies give ev idence for a potential posit ive effect on memory by an acute dose of caffeine in rodents, they dont pr ovide much insight into the effect a longterm treatment of caffeine might have on memory and learning in rodents. Furthermore, it is also difficult to tell if improved performance in these cognitive tasks following acute doses of caffeine is simply related to increased alertness in these mice and not an increase in memory or learning. For instance, caffein e doses of 2.5 and 5 mg/kg had no effect on


41 avoidance learning of mice in the shuttle-box avoidance test while caffeine doses of 10 mg/kg actually impaired their performance (Sansone et al., 1994). Similar results were found by Izquierdo et al (1979), who found mi ce administered 29.9 mg/kg of caffeine were significantly impaired in their memory re tention in the passive avoidance task. It is likely alertness is less of a component in mouse performance in the avoidance tasks, because these tasks utilize a conditioned response rather than the reliance on spatial cues for orientation in the Morris water maze. Las tly, a study using wild type rats found that acute doses of caffeine improve tracking pe rformance, indicating improved alertness (Evenden et al., 1993) (see Table 1 fo r a summary of these results). Controlled human studies have produced mixed results. James (1998) reported no effects of caffeine on performance were seen when administered on either an acute or chronic basis. This study did find participan ts were more alert and less tired following acute intake of caffeine, but felt less alert following chronic exposure to it. A study that stratified the dose-dependent effects of caffe ine found caffeine administered at a lower dose (250 mg) produced pleasant subjective fee lings and positive effects on performance whereas a higher dose (500 mg) produced unplea sant subjective feelings and decreased performance (Kaplan et al., 1997). Differen ces in performance were also reported between studies using low dosages of caffein e. A low dose of caffeine (100mg) failed to have an effect on short or long-term memo ry retrieval in middle to elderly men and women (Schmitt et al., 2003). A study util izing a CANTAB batte ry found 60 mg of caffeine sped up reaction times in pattern recognition, delayed match to sample, and match to sample visual searches (Durlach et al., 1998). Both of these studies are similar however because they included habitual caffein e users forced to abstain from caffeine for


the study purposes. This is significant because another study reported the caffeine-induced increase in a sustained attention task was abolished when the moderate caffeine users were tested after no longer being deprived of caffeine (Yeomans et al., 2002). The authors go on to theorize that caffeines beneficial affect on memory is mainly due to withdrawal reversal, a likely suggestion supported by information gleamed from animal studies dealing with acute or chronic doses of caffeine. Table 1. Immediate Effects of Moderate Caffeine IntakeRodent Cognition *UnknownImproved reference memory (Morris water maze)Striatal A2AantagonismIncreased anxietyUnknownImproved learning/consolidation (Y water maze task)Mesopontine Cholinergic A1antagonismIncreased cortical EEG activityA1antagonismNeurotransmitter DisinhibitionMesopontine Cholinergic A1antagonismIncreased alertnessGeneral Behavior (Humans and Rodents)UnknownIncreased glucose utilizationGlobal vasculature A2AantagonismCerebral VasoconstrictionReceptor Action Physiological *Improved cognition may be due to increased alertnessTable 1. Immediate Effects of Moderate Caffeine IntakeRodent Cognition *UnknownImproved reference memory (Morris water maze)Striatal A2AantagonismIncreased anxietyUnknownImproved learning/consolidation (Y water maze task)Mesopontine Cholinergic A1antagonismIncreased cortical EEG activityA1antagonismNeurotransmitter DisinhibitionMesopontine Cholinergic A1antagonismIncreased alertnessGeneral Behavior (Humans and Rodents)UnknownIncreased glucose utilizationGlobal vasculature A2AantagonismCerebral VasoconstrictionReceptor Action Physiological *Improved cognition may be due to increased alertness Long-Term Effects of Caffeine Intake Behavioral Effect (Rodents). The locomotor disturbances induced by acute doses of caffeine disappear after chronic caffeine administration. For instance, rats given an acute dose of caffeine show decreased exploratory behavior in the open field task and also show increased anxiety in the plus-maze test; yet after 21 days of chronic caffeine administration both of these disturbances disappear (Bhattacharya et al., 1997). Another study reported that tolerance to A 1 receptor blockade is indicated for the lack of consistent locomotor disturbances after caffeine intake in caffeine tolerant individuals 42


43 (KarczKubicha et al., 2003). Overall, thes e animal studies point to a minimizing or complete loss of the motor effects of caffeine once tolerance develops. The chronic administration of caffeine to rodents leads to adaptive changes that abolish the cognitive changes seen during acute treatment of caffeine as well. Chronic administration of the A 1 selective receptor antagonist CP X for only 9 days resulted in a tolerant adaptation that caused CPX to have no effect on the spatial learning and memory of mice tested in the Morris water maze (Von Lubitz et al., 1993). Chronic caffeine treatment for 15 days also had no effect on prevention of memory loss in trained rats (Molinengo et al., 1994). Prio r studies in wild-type rode nts thus have shown that administration of caffeine or specific ade nosine receptor antagonists for approximately two weeks results in no obvious effects on me mory or learning, a lthough no truly chronic caffeine administration studies (that might rela te to caffeine use in humans) have been done investigating cognitive measures. Behavioral Effect (Humans). Large population studies done to examine the long-term effects of caffeine use on human me mory have been controversial because of the many variables that must be taken into account. For inst ance, Hameleers et al. (2000) found in a cross-sectional study of 1875 adults participating in the MAAS study (the Maastricht Aging Study) and ranging in ag e from 24-81 years that higher habitual caffeine intake is significantly associated with faster response speed and improved longterm memory, yet no differences in shortterm memory were found. In contrast, a longitudinal study of the same MAAS group found no relationship between habitual caffeine intake and verbal memory performa nce (van Boxtel et al., 2003). An earlier study of 9003 British adults reported that aged in dividuals appear to be more receptive to


44 caffeines cognitive boosting affects than younger individuals (Jarvis, 1993), thus suggesting that the inclusion of younger adults in the MAAS studies may have provided the confounding results. Physiological Changes of Long-Term Caffeine Intake The study by Maia and Mendonca (2002) that suggested caffeine intake is associated with a lower risk for AD has driven several studies to investigate caffeines antagonism of specific adenosine receptors in the CNS, the more obvious effect of caffeine intake (DallIgna et al., 2004; DallIgna et al., 2003) These studies fail to fully account for the typical patterns of caffeine consumption in humans however. K nowing that caffeines half-life in humans can be less than 3 hours, it is reasonable to a ssume that the short term effects of caffeine intake only exist for brief durations duri ng the typical day. This makes it difficult to attribute an overall protec tion against AD (which is pr ogressively and continually disrupting the normal brain physiology decades before any symptoms appear) to the immediate blockade of adenosine receptors. It is therefore relevant to look to any long term effects of caffeine as the main agen t granting protection from AD. From this standpoint, caffeines neuroprotec tive effect can be elucidated from several studies indicating an effect of caffeine on adenosin e modulation separate from acute receptor antagonism. These are significant because re lationships between impaired adenosine levels and Alzheimers disease have alre ady been implicated but not completely explored. It is particularly noteworthy that incr eased plasma levels of homocysteine, a known risk factor for AD, are associated with decreased adenosine formation in plasma taken from AD patients (Selley, 2004). Interestingly, increased levels of homocysteine


are associated with normal aging, decreased physical activity, and diets high in animal protein yet deficient in fruits and vegetables (Miner et al., 1997). These are all also risk factors for AD. At the molecular level, high levels of homocysteine interfere with the intracellular production of adenosine by forcing the normal conversion of S-adenosylhomocysteine (SAH) into adenosine and L-homocysteine to occur in reverse, thus sequestering adenosine as SAH inside the cell (Fredholm et al., 1999; Fig.2). Indeed, high levels of SAH have recently been identified in the brains of AD patients, and elevated levels of SAH in AD patients have been shown to inhibit important methyltransferases in the brain (Kennedy et al., 2004). Furthermore, the enzyme S-adenosylmethionine (SAM), which is involved in the continuous conversion of methionine to homocysteine (Miner et al., 1997), is found in decreased concentrations in Alzheimers disease (Morrison et al., 1996; Morris, 2003; Mizrahi et al., 2003). The SAM enzyme has recently been identified as a methyl donor to a promoter site on the PS1 gene resulting in decreased PS1 expression (Scarpa et al., 2003), and also is involved in the down-regulation of BACE (-secretase) (Fuso et al., 2005) (Fig. 2). S-adenosylhomocysteine(SAH)Figure 2. Impact of elevated levels of homocysteine on SAM in ADHomocysteine + Adenosine MethionineS-adenosylmethionine (SAM) PS1 and BACE expression AProduction S-adenosylhomocysteine(SAH)Figure 2. Impact of elevated levels of homocysteine on SAM in ADHomocysteine + Adenosine MethionineS-adenosylmethionine (SAM) PS1 and BACE expression AProduction 45


46 If high levels of homocysteine are pres ent, either because of diet, vitamin B 12 /folate deficiency, or advanci ng age; then it is likely that SAM methylation of PS1 and BACE is blocked by the increased concentra tions of SAH. Thus, decreasing methylation by SAM may result in a rise in expression of the PS1 and BACE genes, leading to increases in A production. It has also been found through in vitro studies that a folic acid deficiency and elevated homocysteine levels disrupt DNA repair in hippocampal neurons, which sensitizes them to the toxi c affects of amyoid accu mulation (Kruman et al., 2002). This information implies that incr eased homocysteine levels may play both a direct and indirect role in Alzheimers diseas e, and represents a potential target for the long-term effects of caffe ine intake (see Figure 3). Interestingly, a recent in vivo experiment found extracellular levels of adenosine were elevated 8 hours after caffeine administ ration in rats (Conla y et al., 1997). The authors proposed this increase may be a resu lt of adenosine receptor blockade. However, caffeine is the downstream precursor of a biosyn thetic process in tea leaves that begins with SAM (Koshiishi et al., 2001). Thus, caffe ine intake may elevate SAM levels when it is degraded. It is also possibl e that caffeines blockade of A 2A receptors on astrocytes leads to inhibition of COMT (catechol-O-methyltransferase), the enzyme responsible for conversion of SAM to SAH. The elevati on of SAM would lead to the decreased expression of the genes that would otherwise lead to AD (see Figure 3) and might be evident by increased adenos ine levels as SAH was hydrolyzed to adenosine and homocysteine.


S-adenosylhomocysteine(SAH)Figure 3. Proposed impact of caffeine on the SAM/SAH cycle in ADHomocysteine + Adenosine MethionineS-adenosylmethionine (SAM) PS1 and BACE expression AProduction CaffeineCOMT -1-2-S-adenosylhomocysteine(SAH)Figure 3. Proposed impact of caffeine on the SAM/SAH cycle in ADHomocysteine + Adenosine MethionineS-adenosylmethionine (SAM) PS1 and BACE expression AProduction CaffeineCOMT -1-1-1-2-2Tolerance to Caffeine The long-term administration of caffeine results in the disappearance of the behavioral side effects associated with caffeine intake (e.g. tolerance). Concordantly, this should be the result of neuromolecular changes because of the fluctuating presence of caffeine and elevation of adenosine levels. Indeed, numerous studies involving rodents have reported that chronic treatment with caffeine, ranging from 4-28 days, results in increased numbers of A 1 receptors in both cortical (Tsutsui et al., 2004; Shi and Daly, 1999; Shi et al., 1993) and hippocampal neurons (Rudolphi et al., 1989; Johansson et al., 1993). The mechanism for this increase remains unresolved, as none of these studies reported changes in mRNA for the A 1 receptor. Many of these studies examined A 2a receptors as well, but no changes were seen in receptor counts or mRNA levels (Shi and Daly, 1999; Johansson et al., 1993; Shi et al., 1993). These findings suggest A 1 receptors are responsible for the development of tolerance to the behavioral and physiological side effects of caffeine. Importantly, this increase in A 1 receptor levels has implications for potentially reversing trends seen in AD as well. Additionally, a 4-day treatment of caffeine to mice also found an approximate 17% increase in the density of cortical L-type 47


48 calcium channels, but the authors failed to suggest a mechanism for this and its implications remain unknown (Shi et al., 1993). The considerable 8 hour delay between ca ffeine administration and increases in adenosine first noticed by Conlay et al. was la ter replicated and traced to the blockade of A 1 receptors (Andresen et al., 1999). The aut hors suggested the time lag between receptor antagonism and increases in adenosine may be caused by new protein synthesis, posttranslational modifications, or reductions in the synthesis of key enzymes responsible for metabolizing adenosine, such as adenosine d eaminase. The authors go on to propose that A 1 receptors monitor extracellular levels of adenosine and when activated, influence adenosine levels by potentially modulating the activity of adenosine deaminase. Therefore the losses in hippocampal A 1 receptors reported in patients with dementia (Deckert et al.., 1998) and Alzheimers dis ease (Ulas et al., 1993) suggests impaired monitoring of adenosine function in AD. If ad enosine is being trapped intracellularly in the form of S-adenosylhomocysteine, then th e reduction of extrace llular adenosine could potentially lead to the reductions in its receptor as well. Given the likely role of APP in axonal transport (Kamal et al., 2001) and accumulating A interferes with fast anterograde and retrograde axonal transport (H iruma et al., 2003), it is also possible to surmise impaired axonal transport may l ead to the decrease in adenosine A 1 receptors seen in AD cases. Diminishment of A 1 receptor function would impair an innate neuronal pathway that monitors Ca 2+ homeostasis, which is known to be disrupted in neurons after long term exposure to the A 1-42 isoform found in AD. As men tioned earlier, NMDA receptors are found in abundance in the hippocampus making neurons in this area particularly


49 sensitive to excitotoxic damage during patholog ical conditions. Therefore, it is likely that the progression of AD leads to decreases in A 1 receptor-mediated protective mechanisms against excitotoxicity in both the cerebral cortex and hippocampus in combination with the toxic affects of A creating a hostile environment fo r neurons in these brain areas. An increase in A 1 receptors following long term trea tment of caffeine thus may reverse this trend restoring A 1 receptors to normal or above physiological levels. Health Risks of Caffeine Intake The impact of moderate caffeine intake on health has been scrutinized through epidemiological studies with no clear detrim ental affects. Caffeines relationship with blood pressure is of particul ar concern, yet most studies report no association between caffeine consumption and blood pressure in tolerant individu als (Chou and Benowitz, 1994; Robertson et al., 1984; Bertrand et al., 1978). Other studies have found an association between caffeine and changes in blood pressure, but these findings are considered controversial because they fail to distinguish between infrequent caffeine users and habitual caffeine users (Rachima-Maoz et al., 1998). It has also been suggested that hypertensive patients are more prone to the pressor effects of caffeine, although this sensitivity was not found in a case study (R obertson et al., 1984). Moderate caffeine intake is known to raise blood pressure in caffeine-nave subj ects and in subjects abstaining from caffeine long enough to lose their tolerance to caffeine, yet this pressor affect is only in the range of 2-3 mm Hg and this affect is lost within 24 hours after tolerance appears (Myers, 2004). High intake of caffeine also can lead to tachycardia, heart palpitations, and a small decrease in heart rate, but again these effects are minimized after tolerances develops within days (Fredholm et al., 1999).


50 The Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure expressed their view that re striction of caffeine intake is not a recommended procedur e on reducing blood pressure (Chobanian et al., 2003), yet some doctors continue to recommend the el derly or people alrea dy prone to hypertension avoid caffeine. If caffeine does have neuropr otective properties, it would be prudent instead to recommend a moderate, sustained ca ffeine intake throughout life to reduce any of the cardiovascular effects discussed previ ously that could appear if tolerance to caffeine is lost during any abstinence from the drug. Caffeine, as well as other methylxa nthines, is a well known diuretic and natriuretic that inhibits proximal t ubular reabsorption by antagonizing renal A 1 receptors (Rieg et al., 2004). Finally, caffein e also raises the respirator y rate in the same, but yet not as effective, mechanism as theophyllin e (the common bronchodila tor prescribed for asthmatics) (Benowitz, 1990). Caffeine also ha s weak self-reinforcing properties when compared to the drugs of abuse, such as coca ine or heroine, and is not described as an addictive substance in the Diagnostic and statistical manual of mental disorders (4 th ed.). Although symptoms such as headache and fatigue may accompany caffeine withdrawal when caffeine intake is reduced significantly (Griffiths et al., 1990), none of the side effects of prolonged, moderate caffeine intake have been linked conclusively to any adverse health conditions. Summary Although blockade of A 1 and A 2A adenosine receptors produces some potential immediate neuroprotective effects, it is lik ely that caffeines bi ggest contribution to neuroprotection in Alzheimers disease may be the reinforcement of SAM methylation of


51 PS1 and BACE. Given the nature of moderate caffeine consumption and its metabolism, the short-term neuroprotective effects of caffeine would cer tainly not exist long enough to have any lasting contribution against th e lifelong onslaught of AD. Although it has yet to be determined exactly how caffeine raises SAM levels, several possibilities exist. First, caffeine is a xanthine derivative of SA M, and once metabolized may provide extra substrate for the synthesis of SAM, leading to the increased formation of this compound. Caffeine may also inhibit COMT activa tion, which would decrease astrocytic homocysteine synthesis (Huang et al., 2005) and reduce SAM conversion to SAH. Regardless, the chronic use of caffeine gene rates a rise in SAM that likely leads to diminished expression of genes that are known to be overexpressed in familial Alzheimers disease and are likely askew in sporadic AD, resulting in decreased production of A and prevention of the corres ponding changes in neuropathology. Ideally, changes in expression of these gene s is minimal but sufficient enough to slow down A production thus avoiding many of the negative side effects of other therapeutics designed to minimize expression of these gene s. It is also possi ble that diminished production of A allows for innate protective mechanisms in the body to equilibrate the clearance and production of Caffeine may also be effective in reversing the decreases in hippocampal A 1 receptors seen in AD and aging th rough mechanisms that develop as tolerance in response to chronic caffeine intake, ultimately restoring the endogenous protective pathway against alterations in Ca 2+ or excitotoxicity.


52 IV. Specific Aims Overexpression of the hAPP gene with the Swedish double mutation in APP sw mice results in impaired reference/learning memory and impaired working memory by 89 months of age that closely mimics the cogni tive decline seen in Alzheimers patients. These behavioral impairments are cert ainly linked to the appearance of A species, but other pathological changes must also be taken into consideration as well when evaluating the full scope of the disease. The decrease in A 1 adenosine receptors found in the brains of late Alzheimers patients may be a marker for other important disruptions in the brain, especially considering adenosines relationship with homocysteine/SAH/SAM and its potential indirect impact on en zymes reported to decrease e xpression of genes involved in AD. SAM-related changes have yet to be ex amined in animal models for Alzheimers disease, so it is therefore relevant to s ee if there are any differences between APP sw and non-transgenic mice. Caffeine use has been implicated with the occurrence of Alzheimers disease, with moderate caffeine intake decreasing the risk for the disease. Additionally, caffeine use has been shown to increas e cortical and hippocampal A 1 receptors and raise extracellular adenosine levels in blood plasma. If this occurs in the brain, it may further support the link between impaired SAM functi on and Alzheimers disease. I propose that APP sw mice will have decreased cortical and hippocampal A 1 receptors and subsequently, will also have reduced extracellular brain adenosine and S-adenosylmethionine levels in


53 the brain. These APP sw mice will also be cognitively impaired because of increased levels of soluble A Furthermore, long-term administra tion of a moderate caffeine dose to APP sw mice will reverse these effects, ultimatel y reducing amyloid load in these mice and restoring cognitive function. The specific aims of my research proposal are: To behaviorally char acterize a group of APP sw and Tgcontrol mice in a full battery of sensorimotor and cognitive task s to determine the degree of behavioral impairment these mice develop at 8 months of age. To determine the behavioral effect a 4 month treatment of oral caffeine (beginning at 4 months of age) has on a similar group of age-matched 8 month old APP sw transgenic mice. To explore any differences in SAM and adenosine between non-transgenic and APP sw mice. To evaluate the relationship between longterm caffeine administration in a mouse model of Alzheimers disease and any changes in amyloidogenic processing in the CNS as a result of caffeine treatment. To determine the relationships between brain A levels, brain adenosine receptor levels, and cognitive performance in APP sw and caffeine-treated APP sw transgenic mice.


54 V. Materials and Methods Animals 57 mice were included in this study. E ach mouse had a mixed background of 56.25% C57, 12.5% B6, 18.75% SJL, and 12.5% Swi ss-Webster. All of these mice were derived from a cross between P (parental generation) heterozygous male mice carrying the mutant APP K670N, M671L gene (APP sw ) with F1 PS1 (transgenic line 6.2) female mice to obtain an F2 generation c onsisting of APP/PS1, APP sw PS1, and non-transgenic mice. After weaning, the mice were genotyped with only APP sw and non-transgenic mice selected for behavioral testing and/or caffeine administration. These mice were then group housed in cages with rodent chow and water or caffeinated water ad libum All mice were maintained in a 10 hour dark and 14 hour light cycle at all times, and all behavioral testing was perfor med during the light cycle. Effects of Long-Term Caffeine Administration in Young Adult APP sw Mice General Protocol A total of 41 single transgenic APP sw mice and 16 non-transgen ic (NT) littermates were randomly selected for this study. 14 APP sw mice were randomly selected from the APP sw group and were administer ed caffeine treatment in dr inking water beginning at 4 months of age. The remaining 27 transg enic mice, as well as 16 non-transgenic littermates, were provided normal water to se rve as Tg and NT controls. All mice were removed from group housing and moved to sing le housing two weeks before behavioral


testing began. Behavioral testing began at 8 months of age (four months into caffeine treatment) and consisted of a 6-week battery that was composed of 3 sensorimotor-based tasks, one anxiety-based task, and five cognitive-based tasks. These tasks were performed in the following order: open field activity, balance beam, string agility, Y-maze, elevated plus maze, Morris water maze, circular platform, platform recognition, and radial arm water maze. All mice were euthanized following completion of behavioral testing and brains were removed for further analysis. This analysis consisted of quantification of brain soluble/insoluble A levels, determination of both and -CTFs, and assessment of brain adenosine receptor densities. Fig. 1 depicts a timeline for this study. Mice were weighed every two weeks to ensure no changes in weight occurred from the caffeine treatment. Study Timeline 6-Week Behavioral Battery Oral Administration of Caffeine Animals Sacrificed 55 Fig 4. General protocol time line for long-term caffeine administration study. Animals Born 12 4 0 8 Months


56 Caffeine Treatment At 4 months of age, 14 APP sw group housed mice were given ad libum access to only water with 0.3 mg/mL caffeine ( Sigma) dissolved in it. On average, mice drink 5 mL per day, giving a daily dose of 1.5 mg of ca ffeine to each mouse. Given that metabolic rate (MR) =Mass the MR of mice (average weight = 0.025kg) is 7.2 x greater than humans (average weight = 68kg). Thus, a 1.5 mg daily dose in a mouse is equivalent to an approximately 500 mg daily caffeine intake (~5 cups of coffee) by a human. The caffeinated water was changed two times a week to ensure caffeine remained fully dissolved at the appropriat e concentration. Control APP sw and NT mice were given ad libum access to untreated tap water that was also changed twice weekly to ensure freshness. Mice were kept under these conditions for 3 mont hs, at which point they were separated into single housing, two weeks before behavioral te sting began (Fig. 1). Caffeine treatment was continued throughout th is time and during behavioral testing. The weights of the animals were monitored thr oughout this study to ensure no significant weight reductions occurred. Behavioral Assessment Over a 6-week time course, mice were beha viorally tested for characterization of their sensorimotor, anxiolytic, and cognitive functions utilizing the following tasks and in the order described: Open-Field Task. This test assessed explorat ory behavior and activity by placing mice into an unfamiliar open black box (81 x 81 cm) with 28.5-cm walls. The bottom of the box was marked by 4 horizontal an d 4 vertical lines, dividing the surface into 16 squares. The task consists of a single trial, wherein a single mouse is placed in the


57 center of the field and allowed to explore fo r 5 minutes. During this period, each line crossing was recorded. Before each trial, the surface of the box was sprayed with a diluted vinegar solution to erase any scent cues. Balance Beam Task. The balance beam consists of a 1.1-cm-wide dowel beam suspended 43 cm above a padded surface. Fl anking each end of th e 51-cm-long dowel are 14 x 10.2 cm platforms. Each animal was placed perpendicular to the dowel at the center of the beam and released for an interval of 60 seconds. The duration the animal was able to stay on the balance beam was recorded. If the animal remained on the balance beam for the full time and/or escaped to one of the platforms, the maximum score of 60 seconds was recorded. Each mouses balan ce and general motor function was evaluated by subjecting the animals to 3 trials, with the overall average indicating the best approximation of the mouses performance. String-Suspension Task. This task is an additional sensorimotor test used to characterize the agility and grip strength of mice. Animals we re allowed to grip the string with only their forepaws and released for a single trial of 60 seconds. Each animal was scored on a 0-5 rating system (0=animal was una ble to stay on the string; 1=animal was able to hang onto the string for 60 s by only two forepaws; 2=was given if the animal was able to hang onto the string by two forepaws and one hind limb; 3= animal remained on the string for 60 s and gripped the string by two forepaws and both hindpaws, 4=animal was able to grip the string with four paws and its tail; 5=was given if the mouse escaped from the string to one of the support columns. Y-Maze Task. This task was used to asse ss basic mnemonic processing (by spontaneous percent alternat ion) and exploratory activity (by total number of arm


58 choices) of mice placed into a black Y-maze. The arms of this maze were 21 cm-long and 4 cm-wide with 40 cm-high walls. Each mouse was placed in the center of the maze, facing the arm designated number two and allowed one five minute trial of free exploration of the three alleys in the maze. The number of total arm choices and sequence of arm choices were recorde d. Alternation percentage is defined by the proportion of arm choices that differ from the last two choi ces. For instance, if a mouse selected the following sequence of arm choices (1,2,3,1,3,1,2,1) the total number of alternation opportunities would be six (total entries minus two) and the percentage alternation would be 50% (three of six). Before each trial, the in terior of the maze was spray with a diluted vinegar solution to er ase any scent cues. Elevated Plus-Maze. The elevated (+)-maze was used to assess anxiety in mice. The task has four arms (30 x 5 cm) attached to a 5 x 5 cm central area, all made of plywood and painted black. Two opposite facing arms were unenclosed and open to the surrounding environment. The other two opposite facing arms were enclosed by black aluminum sheet walls (15-cm height). This entire structure sits on a wooden pedestal, elevated 82 cm above the floor level. Each m ouse was placed into the center area facing a closed arm and allowed to explore the pl us-maze for a single five minute trial. The number of closed and open arm entries, a nd the amount of time spent in open arms was recorded. Before each trial, the maze was clean ed with a diluted vine gar solution to erase any scent cues. Morris Water Maze. This water-based task was used to evaluate spatial reference learning/memory of the mice. A 100cm circular inflatab le pool was divided into four equal quadrants by black lines draw n on the floor of the pool; an indiscernible


59 9-cm platform was submerged 1.5 cm belo w the waters surface in quadrant two. The environment surrounding the pool was decorated with eye-catching visual cues to aid the mice in orientating themselves with respect to the pool. Each mouse was subjected to four trials a day over a 10 day period. E ach trial began by placing the mouse into a different quadrant and allowi ng it to swim freely for a ma ximum of 60 seconds. The same quadrant start pattern was used each day. Af ter swimming to the platform (or being guided to the platform if the mouse was unable to locate the platform after 60 s), the animal was allowed to remain on the platform for 30 s before starting the next trial. The latency for each animal to locate the platform in all four trials and the average for all trials was recorded. After the tenth day of acquisition testing, a 60 s probe trial was performed the following day to determine memo ry retention. For this single trial, the submerged platform was removed and each mouse was placed into the quadrant opposite to the quadrant that formerly contained the platform in acquisition testing. The animals swim path and number of annulus crossings we re recorded on videotape; the percent of time spent in each quadrant, as well as aver age swim speed, were determined from these videotapes. Circular Platform Task. This task tests reference learning/memory by placing the mice in a curtain-enshrouded 69-cm circular platform with 16 holes (4.5 cm diameter) equally placed 1.3 cm from the outside edge. The holes designated 4, 8, 12, and 16 allowed the placement of an escape box direc tly underneath the hole. Two-dimensional cues were placed on the inside walls of the en closing curtain, as well as on the platforms walls. The aversive light and wind stimuli used to motivate mice to escape the platform was provided by two 150-W lamps placed 76 cm above the platform and a high-speed fan


60 placed 15 cm above the platform. The first day of the circular platform task consisted of shaping (wherein mice were placed into the center of the platform and gently guided to the location of their escape plat form), with the following eigh t days designated for actual testing. The mice were subjected to a single tr ial each day that consisted of placing each mouse in the center of the platform and allowing it a 5 minute maximum to locate and enter the escape box. During the testing period, the total number of errors (head pokes into non-escape holes) and the latency to escape from the pl atform were recorded. The surface of the circular platform was cleaned af ter each trial with a di lute vinegar solution and the escape box location was moved to a di fferent location after each mouses trial to control for any scent cues (although the box remained in the same location for each particular animal over the eight days of testing). Platform Recognition Task. This water-based task characterizes th e ability of the mice to escape to a visible platform in changing locations placed in the same 100-cm inflatable water pool used in the Morris water maze. A delayed latency to escape indicates a potential deficit in visual acuity or more likely, an impaired ability to switch strategies from a spatial cued strategy used in the Morris water maze to a search/recognition strategy needed for this te st. In this task, a 9-cm circular escape platform was elevated 0.8 cm above the surface of the water with an affixed 10 x 40 cm black and white visual cue to clearly mark it as the escape platform. Performance was evaluated over four days of testing, with 4 tr ials per day. Each day, the mice were placed in the pool at the same location for each trial, but the platform was moved successively to a new quadrant for each of the f our trials. If a mouse was unabl e to locate the platform in the 60 s provided, it was gently guided to the platform. A 30 s rest period was given to


61 each mouse on the platform. Latency to find the platform was recorded over all 4 trials each day, which were averaged for statistical evaluation. Radial Arm Water Maze (RAWM). This final water-based task requires the mice to have intact working (short-term) memory and is the most stringent for determining cognitive deficiencies. The RAWM maze inco rporated the same 100-cm circular pool used previously, but also used an aluminum in sert that creates 6 e qual-sized radial arms surrounding a central open swimming area 40 cm in diameter. Each arm was 30.5 cm long and 19 cm wide, and a transparent 9 cm ci rcular platform that rests 1.5 cm below the waters surface was placed near the end of th e randomly assigned goal arm of the maze for each day. The same spatial cues used in the Morris water maze were provided on the walls surrounding the RAWM task throughout testing. Each day of RAWM testing consisted of four acquisition trials and one memory retention tria l. RAWM testing was conducted over nine successive days. For each acquisition trial, a mouse was placed in the water at the entrance of a novel start arm of the maze for that day facing the central swimming area. This start arm was never th e same arm that contained the submerged escape platform, and the start arm sequences and goal arm location were semi-randomly selected each day. The mouse was then allowed to navigate the maze for 60 seconds. During this 60 s, if the mouse chose the wrong arm (that did not contain the escape platform) it was gently guided back to the start arm to renew navigating the maze and an error was recorded. If a mouse fails to select an arm within 20 seconds, it was gently guided back to the start arm and an error wa s recorded. If the animal failed to find the platform at the end of 60 s, it was gently guided to it. Once locating the platform, the mice were allowed a 30 s rest period. The latency to escape the maze and the number of


62 wrong arm choices were recorded over all four successive acquisition trials. Upon completion of the fourth acquisition trial, the mice were returned to their home cage for a 30 minute interval before being re turned to the pool for the fi fth and final trial of the day, the memory retention trial. The last trial of th e four successive trials (trial 4, T4) and the 30-minute delayed retention trial (trial 5, T5) are considered measures of working memory. Any mouse that did not make at leas t 3 choices during a tria l and were unable to locate the escape platform had a penalty a ssessed for that trial. This penalty was calculated by averaging trial one errors for the first three days of testing for animals which could not find the escape platform but made more than three arm choices during those trials. For this study, the penalty error assessed was 7.1. Brain Collection Immediately following completion of behavioral testing, all mice were anesthetized with Nembutal (1mg/10 gm body weight), then pericardially perfused with 0.9% saline. Brains were then removed and spl it into halves by a si ngle mid-saggital cut. The left hemisphere was fixed in 4% para formaldehyde for 24 hours at 4C, followed by graded sucrose solutions (10, 20, and 30% (w /v) sucrose in 0.1 x Sorensons phosphate buffer). The right hemisphere was dissected out into the following areas: cerebellum, anterior and posterior corte x, striatum, and hippocampus. Th ese areas were immediately frozen on dry ice and stored at -80C. For ha lf the animals in each group, the right frontal cortex, striatum, and hippocampus were used to measure the densities of adenosine receptors. For the remaining half of animals in each group, the right hippocampus was used to determine levels of soluble/insoluble and combined right frontal + posterior cortex was used for analysis of and -carboxyl terminal fragments.


63 Analysis of and CTFs (carboxyl terminal fragments) This procedure was performed by Dr. Jun Tan and Kavon Rezai-zadeh. Frontal and posterior cortices were thawed, combin ed, and placed in ice-cold lysis buffer (20 mM Tris, pH7.5, 150 mM NaCl, 1 mM EDTA 1 mM EGTA, 1% v/v Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM -glycerolphosphate, 1 mM Na 3 VO 4 1 g/mL leupeptin) with 1 mM PMSF. Brain tissues were then sonicated on ice for approximately 3 mins, cooled on ice for 15 mins, and then centrifuged at 15,000 rpm for 15 mins. Following homogenization, ali quots corresponding to 50 g of total protein were electrophoretically separated us ing 16.5% Tris-tricine gels Electrophoresed proteins were then transferred to PVDF me mbranes (Bio-Rad), washed in dH 2 O, and blocked for 1 hr at room temperature in Tris-buffere d saline (TBS; Bio-Rad) containing 5% (w/v) non-fat dry milk. After blocking, membrane s were hybridized for 1 hr at room temperature with primary antibody [APPcarboxyl-terminal antibody 369(1:1000)]. Membranes were then washed 3 times for 5 min each in dH 2 O and incubated for 1 hr at room temperature with the appropriate HRP-conjugated secondary antibody (1:1,000, Pierce Biotechnology, Inc. Rockfo rd, Illinois). All antibodi es were diluted in TBS containing 5% (w/v) non-fat dry milk. Blot s were developed using the luminol reagent (Pierce Biotechnology). Blot inte nsities were analyzed qualit atively and assessed a ratio of 2:1 or 1:1 depending on their inte nsities in relation to each other ( -CTF vs. -CTF). Adenosine Receptor Densities (Completed by Dr. Edward Jackson) Western Blotting. The tissues were placed in ep pendorf tubes with 0.1mml SDS buffer (50mMTris, ph 7.0, 2% SDS, 10% gl ycerol) containing protease inhibitors ( g/ml antipain, 1 ug/ml aprotinin, 2 ug/ml leupeptin, 1 mg/ml phenylmethylsulfonyl


64 fluoride) and homogenized with a small pl astic pestle in ice. The homogenate was centrifuged at 12,000 rpm at 4C for 10 min, a nd the supernatant wa s recovered. Protein in the supernatant was terminated by the copper bicinchoninic acid method. Laemmli buffer was added to samples, after which they were placed in boiling water for 5 min and then chilled immediately on ice. 15 g protein/well samples were loaded onto a 7.5-10% acrylamide gel and subjected to SDS PAGE us ing the Bio-Rad minigel system. Proteins were then electroblotted onto a polyvinylid ine difluride membrane. The membrane was blocked with 5% milk for 1h and incubated at 4C overnight with the first antibody (antiA1 1:1000 diluted in PBS containing 0.5% Tween 20; anti-A2a 1:1000 diluted in PBS containing 0.5% Tween 20, antibodies were from Santa Cruz). After three washes with PBS containing 0.5% Tween 20, the membrane wa s incubated at room temperature for 1h with HSP-conjugated sec ondary antibody (Amersham) at 1:10,000 dilution. The membrane was exposed to film and the signal s were detected by a supersignal substrate kit (Pierce). RT-PCR. Total RNA was isolated using Triz ol reagent solution (GIBCO). By using the primer sequences listed in Table 2, 0.5 g RNA was reverse transcribed and amplified using a Titanium One-Step RT-PCR K it (Clontech). Each PCR cycle (a total of 30 cycles for A1 and 32 cycles for A2a) consisted of denaturing at 94C for 30 s, annealing at 64C for 30 s, and extension at 72C for 60 s. The products were separated on a 1% agarose gel with EB staining and the signal was detected by UV.


65 Table 2. Mouse adenosine recep tor PCR primers and cDNA sizes Receptor Accession No. Primer Nucleotides Sequence 5-3 Product Size A1 NM_009629 Forward 659 TAGGGCAACGCCTTTGGGAC 849 Reverse 1507 ATGGGTGTCAGGCCTACCAC A2A NM_009630 Forward 177 GCCATCACCATCAGCACTGG 734 Reverse 910 TCAGGACGTGGGTTCGGATG The band densities were quantitatively measured using Scion-image software. Background signals were obtained in each lane and subtracted from the band densities to correct for the background signal. Effects of Caffeine Administration in Aged APP sw Mice General Protocol At 17 months of age, 8 APP sw mice and 3 non-transgenic (NT) littermates were randomly selected to determine the effect of caffeine administrati on on both extracellular brain levels of adenosine a nd soluble/insoluble brain A levels. Four APP sw group-housed mice were given ad libum access to water with 0.3 mg/mL caffeine (Sigma ) dissolved in it, as per the long-term study. Four APP sw mice and three NT group-housed mice were not administ ered caffeine in their drinking water and thus served as controls. Mi ce were kept under these conditions for 18 days, at which point they were euthanized (Fig. 5) and brains were removed for quantification of extracellular brain adenosine levels, soluble/insoluble A levels, -secretase activity, and S-adenosyl methionine (SAM) levels. The weights of the animals were monitored throughout this study to ensure no signi ficant weight reductions occurred.


Study Timeline Oral Administration of Caffeine Animals Sacrificed 66 Fig. 5. General protocol time line for caffeine administration in aged mice. Brain Collection Immediately following caffeine treatment, all mice were anesthetized with Nembutal (1mg/10 gm body weight). Each mouses skullcap was surgically removed and the brain was quickly excised and split into halves by a single mid-saggital cut. The left hemisphere was rapidly snap-frozen in liquid nitrogen and reserved for measurement of extracellular brain adenosine levels through HPLC. This technique was done quickly to protect the tissue from ischemic conditions which would disrupt any adenosine measurements. The right hemisphere was dissected out into the following areas: cerebellum, anterior and posterior cortex, striatum, and hippocampus. These areas were immediately frozen in dry ice and stored at -80C. Later, the hippocampus was used to measure the levels of soluble and insoluble A through ELISA. Combined anterior and posterior cortices were later analyzed for both -secretase activity and S-adenosyl methionine levels. 18 17 Months 17.5


67 Measurement of Brain Adenosine Levels (Completed by Dr. Edward Jackson) Sample Preparation Half brains were weighe d (50-60 mg) and washed with 500 L cold phosphate-buffer saline (PBS). Tissue was transferred to a centrifuge tube containing 500 L water and then boiled for four minutes to inactivate adenosine deaminase (and any other enzymes present in the sample). The tissue was then homogenized in a power homogenizer and centrifuged at 14,000 rpm for five minutes. The supernatant was drawn off and centrifuge d a second time. The resulting supernatant was loaded onto centrifugal filter devices (Bio max-30, Millipore) and filtered to remove proteins. The filtrate was diluted 1:200 in water and internal standard (adenine 9-D arabinofuranoside) was added to a final concentration of 10 pg/ L. The standard curve was created in water and the samples were analyzed with an LCMS assay. Mass Spectrometry. The assay was developed using a Thermofinnigan HPLC system coupled to a Thermofinnigan LCQ D uo ion trap mass spectrometer equipped with an electrospray ionization source (ESI). The mass spectrometer was operated in the ESI positive ion mode. The analytes were mon itored using single ion monitoring; for adenosine and adenine 9-D arabinofuranoside (internal standard), the m/z was 268. Measurement of -Secretase Activity (Collaborat ion with Kavon Rezai-zadeh) Brain samples consisting of combined fr ontal and posterior cortices were placed in ice-cold lysis buffer (20 mM Tris, pH7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% v/v Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM -glycerolphosphate, 1 mM Na 3 VO 4 1 g/mL leupeptin). Brain tissues were then sonicated on ice for approximately 3 mins, cooled on ice for 15 mins, and then centrifuged at 15,000 rpm for 15 mins. No


68 PMSF was added to these samples as it may disrupt the activity of -secretase. Cell lysates from these samples were then analyzed for -secretase activity using a -secretase kit provided by R&D Systems, Inc.. Followi ng the manufacturers pr otocol, a secretasespecific peptide conjugated to reporter molecules EDANS and DABCYL was added to each cell lysate. Cleavage of this peptide by the -secretase present from each cell lysate results in a fluorescent signa l that is proportional to the level of secretase enzymatic activity. Fluorescent signals were detected using a fluoromentric reader. Protein concentrations were quantifie d to ensure equal protein le vels were present in each sample. SAM (S-adenosyl-methionine) Quantification (Collaboration with Kavon Rezai-zadeh) The same cortical homogenates used for -secretase activity were also used for SAM quantification. SAM was quantified in these samples using a Bridge-It SAdenosyl-Methionine (SAM) Fluorescence Assay provided by Mediomics, LLC. Following the manufacturers protocol, 10 l of each sample were added to a polypropylene Eppendorf tube. 90 l of SAM assay solution was added to each sample and vortexed for ~1 second. Then, 90 l of the resultant mixture from each sample was loaded in a black microplate, covered and incubated in the dark for 30 minutes. The fluorescent intensity was then measured usi ng a fluorescent micropla te reader (settings: excitation ~485 nm; emission ~665 nm). By using the standard curve generated by the fluorescence of known SAM levels, the con centration of SAM in unknown samples was calculated by its fluorescent in tensity. Protein concentrations were quantified to ensure equal protein levels were present in each sample.


69 Determination of Soluble/Insoluble A Levels Tissue Homogenizing/Extraction. To determine the levels of both soluble/insoluble A (1-40) and A (1-42), the hippocampi from mice in each group (6 Tg+caff, 6 Tg+, 1 NT from Study A and 3 Tg+caff, 3 Tg+, 1 NT from Study B) were homogenized in tissue homogenization buffer (THB) (250mM sucrose, 20mM tris-HCl, 1mM EDTA, 1 mM EGTA, and protease in hibitor). 0.4% diethylamine (DEA; in 100 mM NaCl) solution was added to each homogena te and each sample was then centrifuged at 100,000 x g for one hour. The supernatant from each sample was drawn off and stored at -80C for later quant ification of soluble A levels. Formic acid (FA) was then added to the remaining pellet and samples were sonica ted for one minute. Following sonification, samples were spun down at 100,000 x g for one hour. The intermediate phase was drawn off, and added to a FA neutralization so lution. These FA-extracted solutions were designated for quantifying insoluble A levels and stored at -80C. (1-40) and (1-42) ELISA. Sandwich ELISA kits for both A (1-40) and (142) were utilized (Signet labor atories). All extracts were thaw ed and appropriately diluted to ensure results within the standard curve for each kit. Plates were coated with capture antibodies specific to the amino-terminus for either the A peptide (1-40) or (1-42), depending on which kit was being used. Befo re loading the samples, each plate was washed three times in a wash diluent. The standard curve (consisting of samples with known amounts of peptide) and each brain sample extract were run in duplicate and averaged to obtain the most accurate results. After loading the samples, the plate was incubated at 4C overnight and was washed three times the next day to remove unbound peptide. A primary antibody was then added that binds to the carboxy-terminus of the


70 peptide and allowed to incubate for two hours. After washing, a secondary antibody conjugated to horseradish peroxidase was added to bind to the primary antibody and allowed to incubate for another two hours. The plate was again washed and an Ophenylenediamine (OPD)-substrate was added to each well that visualizes the bound A peptide by reacting with horseradish peroxidase This reaction occurred for 45 minutes, at which point a stop solution was added and the plate was read at an absorbance of 490 nm. The optical density (OD) for each sample was measured and compared to the OD of the standard curve, allowing fo r the quantification of bound A (either 1-40 or 1-42 depending on which kit was used) in each well. Statistical Analysis Behavior. Standard one-way ANO VAs were performed using Statistica software to determine inter-group behavioral compar isons between APP mice, caffeine-treated APP mice, and NT mice in the open-field, ba lance beam, Y-maze, elevated plus-maze, Morris water maze acquisition and probe trial, circular platform, platform recognition, and radial arm water maze (RAWM) tasks. Posthoc pair-by-pair differences between groups (planned comparisons) were later de termined by the Fisher LSD test. Two-way repeated measure ANOVAs were also performed to elucidate any behavioral differences in groups in the multi-day tasks across days (platform recognition) and blocks (3-day blocks for RAWM or 2-day blocks for Mo rris water maze acquisition). Mann-Whitney U-tests for non-parametric data were perfor med to compare performance of mice in the string agility task. RAWM swim speed was calculated by averaging overall T4+T5 latency and dividing by the average of overa ll T4+T5 errors. Any animals that were


71 consistently unable to complete any task (e .g., floaters, circlers) were eliminated from statistical analysis. Pathology. Group comparisons for insoluble/soluble A adenosine receptor measurements, secretase activity, SAM levels, and brain extracellula r adenosine levels were done by standard oneway ANOVAs. After ANOVA analysis, post-hoc pair-by-pair differences between groups were analyzed with the Fisher LSD test. Correlation analysis was also performed using Systat software to elucidate any potential relationships between the behavioral, A pathology, and brain adenosine measures. Factor/Discriminant Factor Analysis. Factor analysis (FA) was performed using Systat software to group all behavioral measures into common factors. This allows for the relationship between individual behavioral measures to be determined, and also might indicate potential performance in a task based on previous performance in another task. This FA was performed using all 19 behavi oral measures from all groups (NT, APP sw and APP sw +Caffeine) in this study. To investigate if a relationship exists between behavioral and A pathological measures, correlation analysis was performed us ing Systat software. Correlation analyses were performed between all 19 behavioral measures and all four A measures. Correlation analyses were also performed between all 19 behavioral measures to determine if any inter-task relationships ex isted. Finally, correlation analysis was also performed to determine the relationship between extracellular brain adenosine levels and the four A measures. Additionally, discriminant factor analys is (DFA) was run using Systat software with two different DFA methods (direct entry and stepwise-f orward) using all 19


72 behavioral measures and also only the 8 c ognitive-based measures that loaded from Factor 1. These DFAs determined if the three groups of mice were be haviorally distinct from one another. The direct entry method used all 19 behavioral measures, or the 8 measures from Factor 1 of FA, while the st epwise-forward method started with 19 or 8 measures, respectively, but selected from thes e measures based on their variance to best differentiate between the three groups. VI. Results Weight Analysis Throughout the study, mice in all three groups were weighed every two weeks to ensure no significant differences occurred from the four month caffeine treatment. Before the start of the caffeine treatment, there were no group differences in weight between the transgenic and non-transgenic mice. Additionally, there were no group differences between the NT Tg, and Tg+Caff mice at the end of behavioral testing when the animals were euthanized. Repeated measures analysis revealed no group by time interaction [F (13,260)=0.66; p=0.89], indicating NT, Tg, and Tg+Caff mice all gained weight at a similar rate during the entire four month period of treatment. BehaviorSensorimotor Evaluation Open Field and Y-Maze Entries. Tg mice exhibited in creased open field activity (Fig. 6a) when compared to NT controls (p <0.05), while activity of Tg+Caff mice was not differ from NT mice at 8 months of ag e. Furthermore, both Tg and Tg+Caff mice


73 made significantly more total arm choices in the Y-maze task (Fig. 8a), another measure for activity/exploration, when compared to NT mice (p<0.02 for both groups). Balance Beam. In balance beam testing, there were no group differences between the mice (Fig. 6b), indi cating that all mice had simila r balance ability and intact general motor function. String Agility. In the string agility test (F ig. 6c), the Tg+Caff mice were less agile than the Tg and NT mice (p<0.05), al though this difference was simply because less Tg+Caff mice escaped the string (score of 5) than the other two groups. As Tg+Caff mice averaged a ranking of 4 in this task, they mana ged to cling to the string with all limbs for the full 60 seconds and did not fall. Theref ore, the difference does not imply a motor impairment in Tg+Caff mice. The treatment -related impairment was task specific and, moreover, did not influence performance in the cognitive-based tasks (see correlation analysis). Elevated Plus-Maze Anxiety. There were no group differences in open and closed arm entries (Fig. 7b and c). Tg mice, but not Tg+Caff mice, spent more time in open arms when compared to NT mice (p<0.05; Fig. 7a). BehaviorCognitive Evaluation Y-Maze Alternation. In Y-maze testing for spontaneous alternation behavior, there were no group differences in percen t alternation (Fig. 8b). Tg mice were unimpaired in this task, thus any protective effect from the long-term administration of caffeine was not possible.


String Agility Ratings 0 1 2 3 4 5 **Tg Tg+Caff NTa) b) c) NT TgTg+Caff Tg Tg+Caff NT Balance Beam Seconds 0 10 20 30 40 Open Field Line Crossings 0 40 80 120 160 200 Figure 6. Open field, balance beam, and string agility performance. Tg mice showed increased activity in the open field task. Tg+Caff mice had decreased performance in the string agility task, but di d not have any general motor impairment. = significantly different from NT group at p<0.05, ** = signi ficantly different from both other groups at p<0.05. 74


a) b) c) TgTg+CaffNT Elevated Plus Maze Time in Open Arms (sec) 010203040506070 Open Arm Entries 012345 Closed Arm Entries 048121620 TgTg+CaffNTTgTg+CaffNT Figure 7. Elevated plus-maze performance. Tg mice spent significantly more time in the open arms than NT mice, while Tg+Caff mice were no different than NT mice. = significantly different from NT group at p<0.05. 75


a) b) Y-Maze Arm Entries 01020304050 NTTgTg+Caff Percent Alternation 020406080 * NTTgTg+Caff Figure 8. Y-maze performance. Both Tg and Tg+Caff mice had significantly more arm entries than NT mice. = significantly different from NT group at p<0.02. 76


77 Morris water maze acquisition. Escape latency data fr om this task were divided into five 2-day blocks. There was no overa ll group effect across all five blocks [F(2,50)=1.69; p=0.19] (Fig. 9). However, a strong group by block interaction was evident [F(8,200)=3.51; p=0.0008], which was clearly due to the inabili ty of Tg mice to improve acquisitional performance af ter the second block of testing. Post hoc analysis of individual blocks revealed NT and Tg+Caff mice showed a strong learning effect in the last two blocks acquisition, resulting in significantly faster escape latencies when compared to Tg mice (Fig. 9). Not surprisi ngly, Tg escape latencies on the last day of testing were also significantly higher than NT and Tg+Caff escape latencies (p<0.5 and p<0.02, respectively. Thus, the long-term admi nistration of caffeine protected against Morris maze acquisitional impairments that would otherwise be present in APP sw mice at this age. Results from the Morris maze probe trial are consistent with the protective effect of caffeine treatment in APP sw mice during acquisition (Fig. 10). Tg mice showed no quadrant preference during this spatial memory retention phase of testing. By contrast, Tg+Caff mice showed an exclusive preference for the quadrant formerly containing the submerged platform (Q2) and spent significan tly more time in this quadrant than Tg mice (p<0.025). NT mice showed only a partial preference for Q2, although their percent time spent in Q2 was not statistically different from that of Tg+Caff mice. Tg+Caff mice also made significantly more annulus crossings than Tg mice (p<0.005; Fig. 10), giving further support for a protective effect of caffeine treatment in spatial memory retention.


Morris Maze AcquisitionBlock 12345Latency (sec) 1020304050 Tg NT Tg+Caf f Figure 9. Morris water maze acquisition latencies. The 10 days of acquisition, as measured by latency to find a submerged platform, are presented in five 2-day blocks. Tg mice were significantly impaired in spatial reference learning in blocks 4 and 5, while Tg+Caff mice performed no different than NT mice. = Tg mice had significantly higher escape latencies than both other groups at p<0.05 or higher level of significance. 78


Annulus Crossings 0123456 NTTgTg+Caff** NTTgTg+Caff Morris Maze Retention % Time in Q2 0102030405060 Q1 Q2 Q3 Q4 Figure 10. Probe trial testing for reference memory retention in the Morris water maze. Tg+Caff mice showed an exclusive preference for the quadrant formerly containing the submerged platform (Q2). This preference is also indicated by the significantly higher number of annulus crossings made by Tg+Caff mice when compared to Tg mice. = significantly greater than all other quadrants at p<0.01 and significantly greater than Q2 for Tg group at p<0.025. ** = significantly different from Tg mice at P<0.005. 79


80 Circular Platform. In this task of referen ce learning/memory, there was an overall groups effect over all eight days of testing [F(2,51)=7.66; P=0.001]. Post hoc analysis indicated that Tg mice had si gnificantly higher escape latencies overall compared with NT and Tg+Caff mice (p <0.005; Fig. 11). Additionally, Tg escape latencies were significantly higher than NT and Tg+Caff escape latencies on the last day of testing (P<0.025 and P<0.001, respectively). Very similar group differences were observed when the number of errors was anal yzed (data not shown). For all animals, there was an overall training (days) e ffect [F(7,357)=37.33; p<0.00001], indicating that all animals collectively improved their perf ormance across days de spite the impaired performance of Tg mice. Thus, the otherwise certain impairment of APP sw mice in the circular platform task, evident in both their escape latencies and error making, was prevented by the long-term administration of caffeine.


Circular PlatformDay 12345678Escape Latency (sec) 050100150200250 NT Tg Tg+Caff NTTgTg+Caff Overall Escape Latency (sec) 020406080100120140 Figure 11. Circular platform performance. Tg mice were significantly slower in locating the escape hole over the eight days of circular platform testing, while Tg+Caff mice performed similarly to NT mice. = Tg mice significantly higher escape latencies than both other groups at p<0.01 or higher level of significance. 81


82 Platform Recognition In platform recognition te sting, an overall groups effect was present [F(2,49)=7.44; p<0.002], with post hoc analysis revealing that Tg mice had significantly higher escape la tencies than NT mice (P<0.0005; Fig. 12). By contrast, Tg+Caff mice performed at the same level as NT mice and nearly had lower overall escape latencies compared to Tg mice (p=0.065) The benefits of caffeine administration were evident even on the second day of testi ng and were profound by the last two days of testing (Fig. 12) in that Tg+Caff mice perfor med identically to NT mice and significantly better than Tg mice on both of those final da ys. A strong group by day interaction was present [F(6,147)=4.14; p<0.001] due to the poor performance of the Tg group relative to the other two groups. The apparent difficultie s that Tg mice have in switching between the spatial (cued) strategy used in Morris wa ter maze to a search/recognition strategy in the platform recognition task are reflected in their high es cape latencies. In sharp contrast, the long-term administration of caffeine protec ted against this strategy switching impairment. RAWM For RAWM statistical analysis, data was evaluated across three 3-day blocks for T1 (semi-randomized initial trial) T4 (final acquisition trial), and T5 (delayed retention trial). The beneficial effects of caffeine on working memory were immediately evident in block 1 of testing (Fig. 13). Tg+C aff mice were similar in performance to NT controls and already better than Tg mi ce (p<0.05) during T4, while Tg mice were impaired during both T4 (p<0.0001) and T5 (p<0.00001). In blocks 2 and 3 of testing, Tg mice had significantly higher escape latencies during working memory trials T4 and


Platform RecognitionDay 1234Latency (sec) 0102030405060 Tg Tg+Caff NT Figure 12. Platform recognition performance over four days of testing. By the third and fourth day of testing, performance of Tg+Caff mice was identical to NT mice and much better than Tg controls. = Tg and Tg+Caff mice significantly worse than NT mice at p<0.05, = Tg mice significantly worse than both NT and Tg+Caff mice at P<0.05 or higher level of significance. 83


T1 T4 T5 Latency (sec) 0102030405060 NT Tg Tg+Caff RAWM T1 T4 T5T1 T4 T5 B1 B2B3 Figure 13. RAWM performance over three 3-day blocks. During the first block of testing, Tg mice had significantly slower escape latencies versus NT mice during T4 and T5 (the trials indicative of working memory) while Tg+Caff mice were not different from NT during T4. In the final two blocks of RAWM, Tg performance was significantly higher in T4 and T5 than both NT and Tg+Caff mice, the later two groups being near identical in performance. = Tg mice significantly worse than both NT and Tg+Caff mice at p<0.05 or higher level of significance. = Tg and Tg+Caff mice significantly different from NT mice at P<0.0001, 84


85 T5 compared to NT and Tg+Caff mice (P <0.05 or higher levels of significance). Tg+Caff mice, by contrast, were identical in working memory performance in comparison to NT mice. A strong group by block interaction was present for T5 [F(4,98)=5.66; p<0.0005], which was clearly du e to the much poorer T5 performance of Tg mice compared to the other two groups. The consistent impairment of untreated APP sw mice in T4 and T5 of RAWM testing in dicate impaired working memory in these mice, whereas the long-term admi nistration of caffeine to APP sw mice offers protection against working memory impairment. The complete protection of working memory provided by caffeine is also manifest in overall performance across a ll three blocks (Fig. 14). In overall T1 performance across all three bl ocks, Tg mice had significantly slower escape latencies in comparison to NT mice (P<0.025) while Tg+Ca ff mice were nearly different from NT mice (p=0.051). Escape latencies in overall T4 and T5 performance showed that Tg mice were substantially impaired when comp ared to NT mice (p<0.00001), while Tg+Caff mice were not different from NT mice and si gnificantly better than Tg mice (p<0.005). Additional support for impaired RAWM worki ng memory in Tg mice was evident in by their increased number of overall T4 and T5 e rrors made in comparison to NT controls (p<0.0001 for both; data not shown). This im pairment was also prevented by the longterm administration of caffeine, as Tg+Ca ff mice performed similarly to NT mice in making significantly lower numbers of overall T4 and T5 errors compared to Tg mice (p<0.0001 and p<0.005, respectively). There we re no group differences in swim speed present in RAWM testing, indicated by th e number of seconds taken per arm choice (latency/error ratio) for T4 and T5 combined.


Overall RAWMT1 T4 T5 Latency (sec) 102030405060 Tg Tg+Caff NT Figure 14. Overall RAWM performance. In working memory trials T4 and T5, Tg+Caff mice performed identical to NT controls and were significantly better than Tg control mice, indicating that the long-term administration of caffeine granted complete protection against working memory impairment in RAWM testing. = Tg mice significantly higher latency than NT mice at P<0.025, = Tg mice significantly worse than both NT and Tg+Caff mice at P<0.005 or higher level of significance. 86


87 Multi-metric Statistical Analysis Factor Analysis. Factor analysis of behavi oral measures was performed to determine the underlying relationships between all of these measur es (Table 3). FA involving all 19 behavioral meas ures resulted in 13 of t hose measures loading on four principal factors, which accounted for 61.2% of the total variance (a measure was considered significant for loading on a fact or if its component loading value exceeded 0.600 for that factor). All measures for RA WM, Morris maze acquisition, and platform recognition loaded heavily in factor 1, thus this factor was considered the primary cognition-based factor. The measures for activ ity and/or having an activity component (open field, Y-maze entries, and circular platform errors) loaded into factor 2, while circular platform latency loaded separately into factor 4 and balance beam loaded separately into factor 5. No tasks loaded into factor 3. Correlation Analysis Correlation analyses were performed using all 19 behavioral measures and including all mice (Tg and NT). There were no correlations between the string agility task and any of the cognitive tasks. Additionally, correlation analysis revealed that the amount of time mice spent in the open arm of the elevated plus maze correlated with an increased latency in multiple tasks (Morris maze, circular platform, platform recognition, and RAWM). Th is may indicate that mice with decreased anxiety are less motivated to escape certain tasks. As expected, there were numerous inter-task correlations between measures in the three water-based. For measures taken from the same task, strong intra-task correla tions were also evident, particularly for RAWM and Morris maze tasks.


Table 3.Factor loadings of behavioral measuresFactorAll 19 behavioral measures1(32.61)RM-T4-FinRM-T4RM-T5-FinRM-T5PR-FinPR-AvgWM-FinWM-Avg2(13.97)OFCP-ErrYM-Ent3(10.10)N/A4(8.247)CP-Lat5(6.37)BB Numbers in bold type indicate percent of total variance explained by a given factor. Abbreviations: BB, balance beam; CP-Err, circular platform overall errors; CP-Lat, circular platform overall latency; OF, open field lines crossed; PR-Avg, Platform recognition overall average; PR-Fin, Platform recognition final day average; RM-T4, radial arm water maze latency overall T4; RM-T5, radial arm water maze latency overall T5; RM-T4-Fin, radial arm water maze latency last block T4; RM-T5-Fin, radial arm water maze latency last block T5; WM-Avg, water maze latency overall average; WM-Fin, water maze latency last day; YM-Ent, Y-maze entries. 88


89 Discriminant Function Analysis. DFA was utilized to determine if multiple behavioral measures could distinguish the three groups (NT, Tg, Tg+Caff) from one another (Table 4). The direct entry DF A method (which includes all behavioral measures) and the stepwise forward DFA me thod (which selects behavioral measures from the total number evaluated based on their contribution to the vari ance) were used in this analysis. Direct entry DFA, using all 19 behavioral measures, was very effective in completely discriminating between all three groups (rank order of performance was NT > Tg+Caff > Tg). Additionally, the stepwise forward DFA (using all 19 behavioral measures) was also very effective in comple tely discriminating between all three groups (P<0.0001). For the stepwise forward DFA, five behavioral measures were retained as providing maximal discriminability: three c ognitive-based measures (RAWM overall T4 latency, Morris maze retention, and circul ar platform overall latency) and two sensorimotor measures (Y-maze entries and elevated platform time in open arms). Additional DFAs were performed using only the eight cognitive-based measures that loaded on factor 1 in FA (see Table 3). Both direct entry and stepwise forward DFAs (using these eight cognitive-based measures) were able to completely discriminate between all three groups. Six measures (three from RAWM, two from platform recognition, and one from Morris maze) provi ding the maximal discrimination between NT, Tg, and Tg+Caff groups. Canonical plots of both the 19 and 8 measure stepwise forward DFAs are shown in Fig. 15. DFAs involving all 19 behavioral measures was slightly better at correctly classifying indi vidual animals into their treatment/genotypic group (84-90% correct) compared to DFAs involving the 8 cognitive measures in factor 1 (75-79%).


Table 4 Summary of discriminant function analysesaMeasuresDirect Entry MethodStepwise-forward methodSignificance% CorrectSignificance% CorrectMeasures RetainedAll 19P<0.000190%P<0.000184%RM-T4WM-RetCP-LatYM-EntEP-TOFactor 1P<0.000179%P<0.0001 75%RM-T4 (Eight cognitive measures)RM-T5RM-T4-FinWM-AvgPR-FinPR-Avga P-values are from Wilks's Post hoc analysis revealed complete discrimination between groups. 90

PAGE 100

Canonical Scores Plots -4 -2 0 2 4 -4 -2 0 2 4 Function 1Function 2NTTg+Caff Tg All 19 MeasuresCanonical Scores Plots -4 -2 0 2 4 -4 -2 0 2 4 Function 1Function 2NTTg+Caff Tg All 19 Measures 8 Cognitive Measures from Factor 1 -4 -3 -2 -1 0 1 2 3Function 1 -4 -3 -2 -1 0 1 2 3Function 2 NTTg+Caff Tg 8 Cognitive Measures from Factor 1 -4 -3 -2 -1 0 1 2 3Function 1 -4 -3 -2 -1 0 1 2 3Function 2 NTTg+Caff Tg Fig. 15. Canonical score plots of stepwise-forward DFAs used to compare the overall cognitive performance of the three mouse groups. Each symbol represents the cognitive performance of one animal graphed from the two linear functions derived in the DFA. In both DFAs, all three groups could be completely distinguished from one another. 91

PAGE 101

92 Neuropathologic/Neurochemical Measures: Study A Adenosine Receptor Measures Adenosine A 1 and A 2A receptor densities from striatum, hippocampus, and frontal cortex from mice in Study A were analyzed using western blotting (Fig. 16). No group differences in adenosine A 1 receptors were seen in striatum or hippocampus, but Tg and Tg+Ca ff mice both had significa ntly higher levels of A 1 receptors in the frontal cortex compared to NT mice (p<0.05). Furthermore, no group differences were seen in adenosine A 2A receptors in striatum or frontal cortex, but Tg mice had significantly higher levels of A 2A receptors in the hippocampus versus NT mice (p<0.05). The mRNA levels for A 1 and A 2A adenosine receptors were also analyzed in the corresponding areas using RT-PCR. This evaluation revealed no significant group differences in A 1 receptor mRNA expression in striatum, hippocampus, or frontal cortex. A 2A receptor mRNA expression was below dete ctable levels in frontal cortex and hippocampus, while in the striatum no group differences were qualitatively found in A 2A receptor expression. (1-40) and (1-42) Analysis Shortly following completion of behavioral testing in Study A at 9 months of age (a nd approximately 4 months into long-term caffeine administration), A analysis by ELISA was performed on hippocampus tissues (Fig. 17). Long-term administration of caffeine resulted in significant reductions of both soluble A 1-40 ( 37%) and insoluble A 1-42 ( 32%). From this data, it is clear that the long-term administration of caffeine has an impact on brain A production and/or clearance.

PAGE 102

A1 A2A Striatum Optical Density 05101520253035 Hippocampus 0102030405060 NTTgTg+Caff* Frontal Cortex 0102030405060 Striatum Optical Density 010203040 Hippocampus 01020304050 Frontal Cortex 010203040 * NTTgTg+CaffNTTgTg+CaffNTTgTg+CaffNTTgTg+CaffNTTgTg+Caff Figure 16. Adenosine A1 and A2A receptor densities in the striatum, hippocampus, and frontal cortex (Study A). The long-term administration of caffeine had no effect on A1 receptors in any areas of the brain examined, although a transgenic effect of increased A1 receptor density in the frontal cortex was observed. A further transgenic effect was seen with A2A receptor densities increased in hippocampus of Tg mice, with caffeine administration reducing this effect below significance. = significantly different from NT at p<0.05. 93

PAGE 103

Tg Tg+Caff pmol/g 0 1 2 3 4 Soluble A 1-42 Tg Tg+Caff pmol/g 0 10 20 30 40 50 60 Soluble A 1-40 Tg Tg+Caf pmol/g 0 100 200 300 400 500 600 Insoluble A 1-40 Tg Tg+Caff pmol/g 0 20 40 60 80 100 120 140 160 Insoluble A 1-42 Tg Tg+Caff pmol/g 0 1 2 3 4 Soluble A 1-42 Tg Tg+Caff pmol/g 0 10 20 30 40 50 60 Soluble A 1-40 Tg Tg+Caff pmol/g 0 100 200 300 400 500 600 Insoluble A 1-40 Tg Tg+Caff pmol/g 0 20 40 60 80 100 120 140 160 Insoluble A 1-42 Figure 17. Quantification of soluble/insoluble A 1-40 and A 1-42 in hippocampus of behaviorally tested APPsw mice (Study A). Caffeine-treated mice had significantly reduced soluble A 1-40 and insoluble A 1-42 levels when compared to untreated Tg mice. Tg+Caff mice significantly lo wer versus Tg mice at p<0.05. 94

PAGE 104

95 Comparison of and CTFs To elucidate any changes in and/or secretase activity in combined frontal and posterior cortex ti ssue from behaviorally-tested mice of Study A, -, and -CTFs (carboxyl-terminal fragments) were analyzed by western blotting (Fig. 18). Qual itative assessment of the resultant gel revealed the same six Tg+Caff mice tested in the A ELISA all had twice the -CTF blot intensity of their -CTF blot intensity (2 to 1 ratio), wher eas the same six Tg control mice from the ELISA had more mixed blot intensities that l eaned towards an even ratio between the two CTFs (1.33 to 1 ratio). Using a non-parametric Mann-Whitney test showed the difference between these group in CTF ratio was sta tistically significant (p<0.025). This data implies that caffeine has an impact on the am yloidogenic processing of APP. In NT mice, cortical levels of CTFs were below the limits of detection. Neuropathologic/Neurochemical Measures: StudyB (1-40) and (1-42) Analysis An 18-day administration of caffeine to aged 17 month old APP sw mice (Study B) resulted in a signi ficant reduction of insoluble A 1-42 ( 30%) in the hippocampus (Fig. 19).

PAGE 105

TgTg+Caff Ratio -CTF to -CTF Ratio Figure 18. Comparison of -CTF to CTFs ratio in APP sw mice following long-term caffeine administration (Study A). Caffeine-treated mice had a significantly greater ratio compared to untreated Tg mice, consistent with a shift in amyloidogenic processing. Tg+Caff mice significantly higher versus Tg mice at p<0.025. 96

PAGE 106

TgTg+Caff pmol/g 050010001500200025003000 Soluble A 1-40 TgTg+Caff pmol/g 05101520253035 Soluble A1-42 TgTg+Caff pmol/g 020004000600080001000012000 Insoluble A1-42 TgTg+Caff pmol/g 01000020000300004000050000 Insoluble A 1-40 Figure 19. Quantification of soluble/insoluble A 1-40 and 1-42 in hippocampus of aged APPsw mice (Study B). Caffeine-treated mice had significantly reduced insoluble A 1-42 levels compared to untreated Tg mice. Tg+Caff mice significantly lower versus Tg mice at p<0.05. 97

PAGE 107

98 Determination of -secretase Activity. To more fully dete rmine if a reduction in -secretase activity is involved in the reduction of amyloidogenic processing after caffeine administration, combined frontal and posterior cortex tissue from the mice in Study B were used to determine enzymatic activity of the -secretase class of proteases using a fluorometric reaction. There were no significant group differences between Tg+Caff and Tg mice (Fig. 20). However, secretase activity in Tg+Caff mice was comparable to NT mice, with both of these groups having slightly reduced secretase activity ( 10%) when compared to untreated Tg mice. It is possible that endogenous mouse secretase activity, unaffected by caffeine treatment, may be present and masking any actual reductions in harmful -secretase activity that may have been provided by caffeine treatment. Regardless, even slight reductions in -secretase activity might have beneficial effects on amyloidoge nic processing, as evident by caffeines ability to reduce hippocampal A levels in these same animals. SAM Quantification. To determine if long-term administration of caffeine has an impact on SAM (S-adenosyl-methionine) levels, combin ed frontal and posterior cortex tissue from the mice in Study B were used in a SAM fluorescence assay ( Mediomics, LLC). There was no significant group differen ce between the Tg+Caff and Tg mice (Fig. 21), yet SAM concentrations were 49% higher in the Tg+Caff mice (7.89 M) than the Tg mice (5.30 M). These results suggest that this analysis shoul d be repeated, but next time using a greater number of samples and a more precise technique in the micromolar ranger.

PAGE 108

Gamma Secretase Activity Fluorescence/g Protein NTTgTg+Caff Figure 20. GammaSecretase activity in cerebral cortex of APP sw mice following caffeine administration (Study B). Caffeine treatment did not significantly affect -secretase activity in our assay. Although it is possible that endogenous mouse secretase activity, unaffected by caffeine treatment, may be present and masking any actual reductions in harmful -secretase activity. 99

PAGE 109

SAM Concentration SAM (M) 024681012 TgTg+Caff Figure 21. SAM (S-adenosyl-methionine) levels in cerebral cortex of APP sw mice following caffeine administration (Study B). There were no significant group differences in SAM concentrations following the caffeine treatment. 100

PAGE 110

101 Extracellular Adenosine Analysis The left brain hemispheres from the aged APP sw mice used in Study B were analyzed fo r adenosine levels by mass spectrometry utilizing an ion-trap Finnegan LC-MS (Fig. 22). Interestingl y, Tg mice had significantly reduced extracellular brain adenosine levels ( 27%) compared to age-matched NT (p<0.005). By contrast, Tg+Ca ff mice had extracellular brain adenosine levels that were similar to NT mice and significantly higher than Tg mice (p<0.05 for Tg vs. Tg+Caff comparison). These results suggest Tg mice ha ve pathologically lowered extracellular adenosine levels and caffeine administra tion restores those levels to normal. Correlation Analyses involving neur ochemical/neuropathologic measures. Correlation analyses were performed betw een all 19 behavioral measures and the four measures of A (e.g., soluble and insoluble A 1-40, A 1-42) for all Tg mice. Other than significant correlations between circular platform impairment and both soluble and insoluble A 1-42, no significant correlations were found between the various forms of quantified A and any of the other behavioral tasks. As expected from the data presented in Figures 16 and 20 involving aged Tg mice, a correlation was found between elevated adenosine levels and decreas ed levels of insoluble A 1-42 (r=0.856; p=0.03) within the hippocampus.

PAGE 111

Extracellular Brain Adenosine ng/mg 020406080100120140160 NTTgTg+Caff** Figure 22. Quantification of extracellular brain adenosine levels in Aged Mice (Study B). The administration of caffeine resulted in a significant elevation of extracellular adenosine levels in Tg+Caff mice to near that of NT mice, while untreated Tg mice had significantly reduced levels. ** = significantly lower than NT (p<0.005) and Tg+Caff mice (p<0.05). 102

PAGE 112

103 VII. Discussion General Summary In the present study, we determined beha vioral and pathological changes in AD transgenic mice mediated by the long-te rm administration of caffeine to APP sw transgenic mice. The 4-month, daily administration of a pproximately 1.5 mg of caffeine to each mouse, equivalent to about five cups of co ffee in humans, resulted in global protection against cognitive impairment. This behavioral protection was associated with reductions in both soluble and insoluble forms of A in hippocampus, supporti ng other reports that cognitive impairment is A dependent. Furthermore, an 18-day administration of caffeine to aged APP sw mice also led to reductions in A and normalized the otherwise decreased extracellular brain adenosine levels in untreated APP sw control mice. Importantly, results from this study establish a direct link between long-t erm caffeine usage and the SAM/SAH cycle, which when potentially disrupt ed in AD, has crucial implications in APP processing and specifically gamma-secretase activity. The complete cognitive protection a nd reductions in AD-like pathology found in this study, together with the recent epidemio logical study indicating a reduced risk of Alzheimers disease from moderate caffe ine use (Maia and Mendonca, 2002), provide a compelling argument for future use of caffeine-based treatments in clinical trials to protect against AD. Given the already widespread use and acceptance of caffeine and the lack of serious side effects associated with long-term intake of caffeine in moderate

PAGE 113

104 amounts, it would be advisable to include a moderate, sustained daily caffeine intake throughout life as a preventative against AD. Behavioral Measures Caffeine administration protected Tg mice against cognitive impairment across tasks testing multiple cognitive domains (reference learning/memory, working memo ry, recognition/identification, strategy switching). The global protection afforded by chronic caffein e administration was not only evident in every task wherein impairment was present in Tg mice, but also incl usive of all cognitive tasks, as discriminant function analysis cl early showed. These wide-ranged cognitive benefits of caffeine administration did not involve significa nt side-effects on sensorimotor function that impacted cogni tive performance because there were no correlations between sensorimotor function and cognitive performance. A discussion of results from each task, in the orde r performed, in presented below. No differences in Y-maze alternations we re seen between the three groups, thus it is likely that the Y-maze task is not strict enough to detect any obvious abnormalities in basic mnemonic processing in 8-9 month old APP sw mice as previous studies have shown (Holcomb et al., 1999; King et al., 2002). Th erefore, with no transgenic impairment present, the long-term caffeine administration to transgenic mice was unable to have any effect in this task. Results from the acquisitional phase of Morris water maze revealed Tg mice had significant impairment in the fourth and fift h blocks, indicating that these mice have impaired spatial learning, as earlier studies have found (Hsiao et al., 1996; Westerman et al., 2002; Arendash et al., 2004). Further impair ment was seen in Tg mice in the memory retention phase of this task, as these mice failed to show any preference for the quadrant

PAGE 114

105 formerly containing the submerged escape platform. In contrast, Tg mice given caffeine had significantly quicker es cape latencies during the acq uisition phase of testing. Additionally, Tg+Caff mice show ed an exclusive preference for the quadrant formerly containing the submerged escape platform, as indicated by their annulus crossings and time spent in this particular quadrant. These results indicate the long-term administration of caffeine protected APP sw mice against spatial/refer ence learning and memory impairment. Tg mice showed additional impairment in the circular platform task, a behavioral task that also relies on spatial/reference le arning and memory. In previous studies, Pompl et al. (1999) reported seven month old APP sw mice were impaired in circular platform performance when the escape holes location wa s changed (e.g., revers al learning). In the present study, Tg mice were worse across all eight days of testing when compared to NT mice in the circular platform task. By contrast, Tg+Caff mice performed identically to NT mice and substantially better than Tg controls in this task. Additionally, Tg mice made significantly more errors than the ot her two groups on the last day of testing, further highlighting the cognitive impairment that the Tg group alone had and the protective affect of caffeine against this impa irment. It is notewor thy that the circular platform impairments evident in the present studys APP sw mice were present without changing the escape wholes location, as was the case for Pompl et al. (1999). If performed after the Morris maze task, the platform recognition task relies heavily on the ability of mice to switch from the cued (spatial) strategy used in Morris water maze to a search/recognition strategy. Difficulties doing this are associated with higher escape latencies. Pr evious behavioral characterizations reported APP sw mice of

PAGE 115

106 similar age were impaired in th is task (Hsiao et al., 1996; Ki ng et al., 2002; Arendash et al., 2004). Tg mice showed an overall impa irment in platform recognition testing compared to NT controls, while Tg mice give n caffeine performed equivalent to NT mice overall. Moreover, on the last day of test ing in this study, Tg mice had significantly higher escape latencies when compared to bo th NT and Tg+Caff mice. Thus, chronic caffeine administration resulted in protection of strategy switching abilities in Tg mice. As the most sensitive of all the cogni tive-based behavioral tasks in our test battery, the RAWM task is pr oficient in elucidating working memory impairment in transgenic mice. An earlier study reported 15 month old APP sw mice were impaired in RAWM testing (Morgan et al., 2000), and Arenda sh et al. (2004) found recently that this impairment is evident as early as 6.5 months of age. Consis tent with both of these prior studies, 8-9 month old Tg mice in the present study were impa ired overall in both T4 and T5 of RAWM (the trials i ndicative of working memory), as evident by significantly higher escape latencies than NT mice. As with all other cognitive tasks wherein the Tg group was impaired, Tg+Caff mice were prot ected against RAWM working memory impairment in performing significantly better than Tg controls and no different from NT mice. The similar performance between NT and Tg+Caff mice in this sensitive task indicates that long-term caffeine treatment grants powerful protection against working memory impairment in these Alzheimers transgenic mice. Consistent with earlier characterizations of this APP sw mouse model (Holcomb et al., 1999; King et al., 2002; Arendash et al., 200 4; Leighty et al., 2004) no differences in overall sensorimotor performance that might indicate a gross motor dysfunction were seen in either Tg or Tg+Caff mice compared to NT controls. The absence of differences

PAGE 116

107 in the elevated plus maze between Tg a nd Tg+Caff mice shows that the long-term administration of caffeine does not result in any anxiety-based side effects that could explain the profound differences in cognitiv e performance between these two groups. These results are consistent with an earli er study reporting the lack of anxiolytic properties associated with long-term caffeine administration to rodents (Bhattacharya et al., 1997). In factor analysis, the loading of RAWM platform recognition, and Morris water maze measures into factor 1 led this factor to be dubbed the primary cognitive-based factor. Furthermore, both the direct en try and the step-wise forward methods of discriminant function analysis were able to completely discriminate between all three groups when using all 19 behavioral measur es, as well as only the cognitively-based measures in factor 1. This indicates that th e long-term administration of caffeine resulted in an overall protection of cognitive function across multiple behavioral measures, although the performance of NT mice across those multiple behavioral measures could still be distinguished from that of Tg+Ca ff mice. Thus, multiple cognitive domains were protected by caffeine treatment in Tg mice. The consistently poor perfor mances of untreated APP sw transgenic mice in four of the five cognitive-based tasks used in the pr esent study indicates that these particular mice exhibit widespread cogni tive deficiencies that negatively impact spatial memory/learning, working memory, and recogniti on/identification in a manner similar to that of early to moderate Alzheimers dis ease in humans. These impairments currently reported are consistent with an earlie r behavioral charac terization of APP sw mice involving the same transgenic colony (Arendash et al., 2004). Even more significant, the

PAGE 117

108 long-term administration of caffeine to Tg mice, presumably started before these mice become cognitively impaired, resulted in a behavioral phenotype similar to that of NT mice. These results indicate caffeines potenti ally beneficial influe nces on the cognitivedestabilizing processe s apparent in Alzheimers diseas e are worth investigating in a clinical manner. Previous studies investigating the beha vioral effects of long-term caffeine administration to normal (wild type) rodents fo rm a consensus that this treatment has no effects on cognition (Von Lubitz et al., 1993; Molinengo et al., 1994), although none of these studies lasted beyond 15 days. Therefor e, it is sufficient to assume that the behavioral improvements seen in the caffein e-treated mice from the present study are a result of behavioral protec tion against over-expression of the mutated APP gene in APP sw mice, rather than a behavioral improvement resulting directly from caffeines immediate effects on cognition. Furthermore, the present study aimed to duplicate moderate dietary caffeine use in humans by placing mice on a truly long-term diet of caffeine administration lasting for four months and c ontinuing throughout beha vioral testing, thus establishing the mice are fully tolerant to caffeine and avoiding the potentially harmful effects of caffeine withdrawal. By ensuring the mice consumed a caffeine dose equivalent to approximately five cups of coffee in humans, the present study is the first of its kind to closely mimic the nature of human caffeine c onsumption that was found to be associated with a reduced risk of Alzheimers diseas e in a retrospective study (Maia and Mendonca, 2002). Neurochemical/Neuropathological Measures Long-term administration of caffeine resulted in no significant effects on A 1 adenosine receptor densities in the

PAGE 118

109 striatum, hippocampus, or frontal cortex of 9 month old, behaviorally-tested APP sw mice. Untreated APP sw mice did have significantly increased A 2A receptors in the hippocampus, a finding that may indicate increased glial activation in this area as A 2A receptors are generally only associated with glial or va scular cells when found within the hippocampus (Fiebach et al., 1996). Interestingly, caffein e administration did reduce these elevated hippocampal A 2A receptor levels to a level that was not statistically different than NT mice, although the reduction was not great e nough to result in a significant difference between Tg and Tg+Caff mice. No differences in A 1 or A 2A mRNA expression were seen following long-term caffeine administration in an y of the groups in any of the brain areas examined, as other labs have reported in nor mal, wild type mice (Johansson et al., 1993; Shi and Daly, 1999). The long-term administration of caffeine resulted in significantly less soluble 40 and insoluble A 1-42 in the hippocampi of caffeine-treated APP sw mice when compared to the age-matched APP sw mice used in this study. Earlie r characterizations of the APP sw mice have indicated their behavioral impairment was A dependent, as age-related increases in amyloid levels (particularl y the soluble isoforms) were found in the hippocampus and/or associate cortices that coin cided with deterioration in cognitive tasks (Chen et al., 2000; Westerman et al., 2002). In the present study, reduced amyloid burden thus appears to be a primary cause for the lack of impairment in cognition-based tasks that the caffeine-treated APP sw mice exhibited. It is also noteworthy that the 18-day administration of caffeine to 17.5 month old APP sw mice also resulted in a significant reduction of insoluble A 1-42 in the hippocampus. This implies that caffeine administration has an effect on amyloid leve ls after they have reached substantial

PAGE 119

110 amounts in aged APP sw mice, indicating long-term caffeine intake may have a treatment role as well as a preventative role in Alzheimers disease. The reductions in soluble A 1-40 and insoluble A 1-42 following caffeine treatment are likely associated with changes in amyl oidogenic processing, as caffeine is too small molecularly to have an effect on bindi ng to and clearing the much larger A species. In cerebral cortical tissues from behaviorally-t ested Tg mice, the increased ratio between and -CTFs mediated by caffeine treatment t hus suggests that caffeine has a direct impact on the secretases involved in amyloidogenic processing. Increases in -CTFs could potentially be construed as a harmful precursor to A generation that might result from increases in -secretase or decreases in -secretase activity (as the -CTF results from -secretase cleavage of APP and is the inte rmediate precursor that when cleaved by -secretase cleavage yields A ). Yet in this case we theorize the increase in -CTFs is a result of decreased -secretase activity. Less -mediated cleavage of -CTFs in the caffeine-treated Tg mice would explai n why they had a greater amount of -CTF present relative to -CTF and would also expl ain the decreases in A found in these mice. When -secretase activity was assayed directly from cortical tissues of aged Tg mice after 18 days of caffeine administrati on, no statistically signi ficant difference in secretase activity was seen between Tg a nd Tg+Caff mice. However, there is the possibility that a larger decrease in -secretase activity was masked by the presence of other endogenous mouse proteins that are unaffected by caffeine treatment yet share secretases preferred substrate used in our assay.

PAGE 120

111 In the present study, aged APP sw mice had significantly reduced brain levels of extracellular adenosine, while the long-term (18 day) administration of caffeine restored adenosine levels in aged APP sw mice to those of NT mi ce. Although mice were give ad libum access to caffeine-treated water, it can generally be assumed that mice consume the most water during nighttime hours and theref ore plasma caffeine levels were likely greatest during the night. In this study, mice were sacrif iced at mid-morning, indicating extracellular brain adenosine levels were significantly elevated hours after peak plasma levels of caffeine have fallen. In a related finding, Conlay et al. (1997) found caffeine administration to rats resulte d in elevated plasma adenos ine levels eigh t hours later. Although it is likely caffeine transiently raises adenosine levels af ter increasing cellular metabolism while increasing vasoconstriction in the brain, this effect should diminish as caffeine is metabolized and its antagonism of adenosine receptors removed. The longer duration of increased adenosine levels (i.e. eight hours after caffeine administration) thus may be a result of increased S-adenosyl-hom ocysteine (SAH) hydrolysis, resulting in the freeing of intracellular adenosine. This potential link between caffeine admi nistration and increased SAH hydrolysis suggests that caffeine may have an imp act on the impaired SAM/SAH cycle in Alzheimers disease (see Figure 2) which has been shown through in vitro studies to affect PS1 and BACE expression (Fuso et al., 2005). Decr eases in SAM (S-adenosylmethionine) concentrations have been repor ted in AD patients (Morrison et al., 1996), while SAH levels are found to be increased in AD patients (Kennedy et al., 2004). In the present study, we found caffeine treatment raised SAM levels by 49% compared to untreated Tg mice. Although th is increase was not statistically signifi cant, possibly due

PAGE 121

112 to either small group sizes (n=3) or euthanas ia at a time well after maximal drinking (e.g., caffeine intake) the data suggest caffeine raises SAM levels, leading to decreased expression of PS1 and BACE and subsequent decreased production of A (see Figure 3). Proposed Mechanism of Caffeine -Mediated Cognitive Improvement The leading theory in Alzheimers re search is the amyloid hypothesis, which proposes that an increased production of -amyloid is responsible for a cascade of NFT formation, oxidative damage, neuroinflammation, and neurodegeneration/neuronal dysfunction. Supporting this theory, transg enic mice overexpressing a hAPP gene with amyloidogenic mutations over-express mutant APP, resulting in production of A and ensuing cognitive impairment. In the present study, long-term caffeine treatment reduced A and protected the mice from cognitive impairment. A previous study investigating the potent ial neurprotective eff ects of caffeine was limited to an in vitro investigation that failed to account for caffeine-mediated physiological changes aside from the pharmol ogical blockade of adenosine receptors (DallIgna et al., 2003). These investigators reported that A 2A receptor blockade by caffeine in cultured rat cerebellum cells resulted in neuroprotection against -amyloid toxicity. Past literature suggests, however, that A 2A receptors are limited to insignificant numbers in cerebellum cells (Fredholm et al., 1999; Fisone et al., 2004). It is also known that cerebellum cells are not norma lly subjected to high levels of -amyloid until possibly the very late stages of advanced Alzheimers disease. If caffeine-mediated blockade of A 2A receptors was neuroprotective, then this would also dictate that caffeine must be present in the bloodstream at all times The epidemiological study by Maia and Mendonca (2002) reported that a caffeine intake of 199 mg/day (approximately two cups

PAGE 122

113 of coffee) reduced the risk of Alzheimers disease, indicating that a sustained presence of caffeine is not required for caffeines prot ective affects on cognitiv e function. In the present study, the natu re of the long-term caffeine tr eatment was designed to closely mimic the normal consumption of caffeine in a typical caffeine user. With this in mind, the in vivo effects of long-term caffeine intake on the cognitive function and pathological features of APP sw mice could be examined through th e present controlled study, which eliminated other potential variables over a protracted longitudinal treatment format. Caffeine is derived from a biosynthetic path way in tea leaves that begins with the precursor S-adenosyl-methionine (SAM) (Koshiis hi et al., 2001). SAM is a major methyl donor in the brain and is invol ved in the methylation status of various genes including PS1 and BACE (Fuso et al., 2005). Under norma l circumstances, SAM is converted to SAH by catechol-O-methyltransferase (COM T) while donating its methyl group and SAH is rapidly hydrolyzed to adenosine and homocysteine. When homocysteine is present in higher concentrations, such as AD, the equilibrium reaction proceeds in the opposite direction and favors the formation of SAH from adenosine and homocysteine. Elevations in SAH could potentially block th e methylation reaction leading to increased expression of PS1 and BACE (Scarpa et al., 2003), and would be ev ident by decreases in extracellular adenosine as it is synthesized into intracellular SAH. Elevated plasma levels of homocysteine are considered a strong independent risk factor for AD (Seshadri et al., 2002), and it was recently reported that AD patients over the age of 60 consumed signi ficantly less dietary vitamin B6 and folate than controls (Mizrahi et al., 2003). Vitamin B6 and folate are both required for the recycling of homocysteine back to methionine, and homo cysteine accumulates when these compounds

PAGE 123

114 are present in insufficient quant ities (Fuso et al., 2005). Dietary influences such as these, as well as other genetic or physical alte rations that might affect homocysteine metabolism, result in accumulation of SAH and the subsequent state of hypomethylation proposed in the genes implicated in AD (Scarpa et al., 2003). It is th erefore relevant to address the link between caffeine, SAM, and SAH in Alzheimers disease. It is currently unknown how caffeine might raise SAM levels as this is the first study to suggest that a caffeine-induced eleva tion in SAM occurs. It is possible caffeine is directly metabolized back to SAM once absorbed into the body, as caffeine is a downstream product of SAM biosynthesis in t ea leaves. Another proposed theory relies on caffeine down-regulating astrocytic metabolism by antagonizing A 2A receptors in these cells, which contain the majority of COMT present in the brain. The potential inactivation of COMT in this manner would lead to a decr ease in transformation of SAM to SAH, promoting the methylation status of PS1 and BACE. On this note, COMT inhibitors have been used to treat depressi on (a symptom of Alzheimers disease) and are reported to decrease the L-dopa induced increases in homocysteine that accompany Parkinsons disease patients (Miller et al., 1997). It was also recently found that direct activation of COMT stimulates homocysteine synthesis in astrocytes, which in turn export homocysteine to neighboring neurons (Huang et al., 2005). A caffeine-mediated inhibition of COMT-induced homocysteine s ynthesis would thus have beneficial implications in a variety of diseases in addition to AD.

PAGE 124

115 Clinical Implications of Caffeine Administration Study Findings and Potential Future Investigations In and of itself, the near-complete protection against cognitive impairment granted by the long-term administration of caffeine to APP sw mice warrants further studies into the relationship between caffeine and Alzheimers disease. Furthermore, the substantial decreases in A following caffeine treatment provide evidence that caffeine does have a positive effect (perhaps indirec tly) on amyloidogenic processing that results in behavioral improvements. The proposed li nk between a caffeine-induced change in the SAM/SAH cycle and the effect this relationship potentially has on SAM methylation of BACE and PS1 activity would be novel in pr oviding a link between a dietary influence and the genetics of Alzheimers disease. In addition to this proposed indirect mechanism for caffeine affecting amyloidogenic pathways, it is possi ble that direct effects of caffeine on neuronal amyloidoge nic pathways are also present. In vitro neuronal culture studies could de termine if such direct affects are, indeed, part of the beneficial mechanisms through which caffeine reduces brain A levels and protects cognitive function. The suggested diminishment of PS1 e xpression following caffeine treatment has potentially powerful repercussions. First, decreased PS1 expression following long-term caffeine use would result in decreases in the pathological amyloidogenic processing found in Alzheimers disease as suggested, but this decrease in expres sion would also not result in a complete loss of function of th e PS1 gene, sidestepping the potential harmful effects that often accompany PS1 inhibitors. Secondly, the 18-day treatment of caffeine used in study B of this investigation also resulted in a substantial decrease in A as well,

PAGE 125

116 indicating that caffeine admini stration may serve as an effective treatment against those already diagnosed with AD. Lastly, the use of caffeine is already widely accepted by the global community and is cost effective. It has also been established that moderate use of caffeine is unaccompanied by harmful side effects as long as caffeine intake is maintained on a daily basis. Thus, it is a safe, naturally-occurring nutraceutic agent that could have significant prophylactic and ther apeutic value against AD, whether taken alone or in combination w ith other AD therapeutics.

PAGE 126

117 References 1. Abbracchio MP, Colombo F, Di Luca M, Zaratin P, Cattabeni F. Adenosine modulates the dopaminergic function in the nigro-striatal system by interacting with striatal dopamine dependent adenylate cyclase. Pharmacol Res Commun 19: 275286. 1987. 2. Acquas E, Tanda G, Di Chiara G. Differential effects of caffeine on dopamine and acetylcholine transmission in brain areas of drug-naive and caffeine-pretreated rats. Neuropsychopharmacology 27: 182-93. 2002. 3. Andresen, BT; Gillespie, DG; Mi, Z; Dubey, RK; and Jackson, EK. Role of adenosine A 1 receptors in modulating extr acellular adenosine levels. J Pharmacol Exp Ther 291: 76-80. 1999. 4. Angelucci ME, Cesario C, Hiroi RH, Rosale n PL, Da Cunha C. Effects of caffeine on learning and memory in rats tested in the Morris water maze. Braz J Med Biol Res 35: 1201-8. 2002. 5. Arendash GW, Lewis J, Leighty RE, McGowan E, Cracchiolo JR, Hutton M, Garcia MF. Multi-metric behavioral comparis on of APPsw and P301L models for Alzheimer's disease: linkage of poorer cognitive performance to tau pathology in forebrain. Brain Res 1012: 29-41. 2004. 6. Arendash GW, King DL, Gordon MN, Morg an D, Hatcher JM, Hope CE, and Diamond DM. Progressive, age-related behavi oral impairments in transgenic mice carrying both mutant amyloid precursor protein and presenilin-1 transgenes. Brain Res 891: 42-53. 2001. 7. Arendash, GW; and King, DL. Intraand intert ask relationships in a behavioral test battery given to Tg2576 tran sgenic mice and controls. Physio Beh 75: 643-652. 2002. 8. Arnaud, MJ. The pharmacology of caffeine. Prog Drug Res 31: 273-313. 1987. 9. Arnaud, MJ. Metabolism of caffeine a nd other components of coffee, in Caffeine, Coffee and Health. pp 43-95, Raven Press, New York. 1993. 10. Benowitz NL. Clinical pharmacology of caffeine. Annu Rev Med 41: 277-88. 1990. 11. Benowitz, NL; Jacob, P; Mayan, H; and De naro, C. Sympathomimetic effects of paraxanthine and caffeine in humans. Clin Pharmacol Ther 58: 684-91. 1995.

PAGE 127

118 12. Benzing, WC; Wujek, JR; Ward, EK; Sha ffer, D; Ashe, KH; Younkin, SG; and Brunden, KR. Evidence for glial-mediated inflammation in aged APPsw transgenic mice. Neurobiol Aging 20: 581-589. 1999. 13. Bertrand CA, Pomper I, Hillman G, Duffy JC, Micheli I. No relation between coffee and blood pressure. N Eng J Med 299: 315-6. 1978. 14. Bhattacharya SK, Satyan KS, Chakrabart i A. Anxiogenic action of caffeine: an experimental study in rats. J Psychopharmacol 11: 219-24. 1997. 15. Blanchard, J and Sawers, SJ. The absolute bioavailability of caffeine in man. Eur J Clin Pharmacol 24: 93-98. 1983a. 16. Blanchard, J and Sawers, SJ. Comparativ e pharmacokinetics of caffeine in young and elderly men. J Pharmacokinet Biopharm 11: 109-26. 1983b. 17. Bonati, M; Latini, R; Tognoni, G, Young, JF; and Garattini, S. Interspecies comparison of in vivo caffeine pharmokine tics in man, monkey, rabbit, rat, and mouse. Drug Metab Rev 15: 1355-1383. 1984. 18. Borchelt DR, Ratovitski T, van Lare J, Lee MK, Gonzales V, Jenkins NA, Copeland NG, Price DL, and Sisodia SS. Accelerated amyloid deposition in the brains of transgenic mice coexpressing mutant presen ilin 1 and amyloid precursor proteins. Neuron 19: 939-945. 1997. 19. Brown SJ, James S, Reddington M, Rich ardson PJ. Both A1 and A2a purine receptors regulate striatal acetylcholine release. J Neurochem 55: 31-8. 1990. 20. Buerger, K; Zinkowski, R; Teipel, SJ; Arai, H; DeBernardi s, J; Kerkman, D; McCulloch, C; Padberg, F; Faltraco, F; Goernitz, A; Tapiola, T; Rapoport, SI; Pirttila, T; Moller, HJ; and Hampel, H. Differentiation of geriatric major depression from Alzheimers disease with CSF tau protein phosphorylated at threonine 231. Am J Psychiatry 160: 376-379. 2003. 21. Calhoun ME, Burgermeister P, Phinney AL, Stalder M, Tolnay M, Wiederhold KH, Abramowski D, Sturchler-Pierrat C, Somm er B, Staufenbiel M, and Jucker M. Neuronal overexpression of mutant amyloid precursor protein results in prominent deposition of cerebrovascular amyoid. Proc Natl Acad Sci USA 96: 14088-14093. 1999. 22. Calhoun ME, Wiederhold KH, Abramowski D, Phinney AL, Probst A, SturchlerPierrat C, Staufenbiel M, Sommer B, and Jucker M. Neuron loss in APP transgenic mice. Nature 395: 755-56. 1998.

PAGE 128

119 23. Castellano C. Effects of caffeine on di scrimination learning, consolidation, and learned behavior in mice. Psychopharmacology 48: 255-260. 1976. 24. Chapman, PF; White, GL; Jones, MW; Coope r-Blacketer, D; Marshall, VJ; Irizarry, M; Younkin, L; Good, MA; Bliss, TV; Hyman, BT; Younkin, SG; and Hsiao, KK. Impaired synaptic plasticity and learning in aged amyloid precursor protein transgenic mice. Nat Neurosci 2:271-276. 1999. 25. Chen, G; Chen, K; Knox, J; Inglis, J; Bernard, A; Martin, S; Justice, A; McConlogue, L; Games, D; Freedman, S; a nd Morris, R. A learning deficit related to age and -amyloid plaques in a mouse model of Alzheimers disease. Nature 408: 975-979. 2000. 26. Chen JF, Xu K, Petzer JP, Staal R, Xu YH, Beilstein M, Sonsalla PK, Castagnoli K, Castagnoli N Jr, Schwarzschild MA. Ne uroprotection by caffeine and A(2A) adenosine receptor inactivation in a model of Parkinson's disease. J Neurosci 21: RC143. 2001. 27. Chobanian AV, Bakris GL, Black HR, Cush man WC, Green LA, Izzo JL Jr, Jones DW, Materson BJ, Oparil S, Wright JT Jr, Roccella EJ; Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. National Heart, Lung, and Blood Institute; Na tional High Blood Pr essure Education Program Coordinating Committee. Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension 42:1206-1252. 2003. 28. Chou TM, Benowitz NL. Caffeine and coffee: effects on health and cardiovascular disease. Comp Biochem Physiol C Pharmacol Toxicol Endocrinol 109: 173-89. 1994. 29. Conlay, LA; Conant, JA; deBros, F; and Wurtman, R. Caffeine alters plasma adenosine levels. Nature 389: 136. 1997. 30. Dall'Igna OP, Porciuncula LO, Souza DO, Cunha RA, Lara DR. Neuroprotection by caffeine and adenosine A2A receptor bloc kade of beta-amyloid neurotoxicity. Br J Pharmacol 138: 1207-9. 2003. 31. Dall'Igna OP, Souza DO, Lara DR. Caffeine as a neuroprotective adenosine receptor antagonist. Ann Pharmacother 38: 717-8. 2004. 32. Deckert J, Abel F, Kunig G, Hartmann J, Senitz D, Maier H, Ransmayr G, Riederer P. Loss of human hippocampal adenosine A1 receptors in dementia: evidence for lack of specificity. Neurosci Lett 244: 1-4. 1998.

PAGE 129

120 33. DeToledo-Morrell, L; Stoub, TR; Bulg akova, M; Wilson, RS; Bennett, DA; Leurgans, S; Wuu, J; and Turner, DA. MRI-derived entorhinal volume is a good predictor of conversion from MCI to AD. Neurobiol Aging 9: 1197-1203. 2004. 34. Dixon AK, Widdowson L, Richardson PJ. De sensitisation of the adenosine A1 receptor by the A2A receptor in the rat striatum. J Neurochem 69: 315-21. 1997. 35. Dodart, J; Meziane, H; Mathis, C; Ungerer, A; Bales, K; and Paul, S. Behavioral disturbances in transgenic mice overexpressing the V717F -amyloid precursor protein. Behav Neurosci 113: 982-990. 1999. 36. Dodart, JC; Mathis, C; Saura, J; Ba les, K; Paul, S; and Ungerer, A. Neuroanatomical abnormalities in behaviorally characterized APP V717F transgenic mice. Neurobiol dis 7: 71-85. 2000. 37. Dodart, J; Mathis, C; Bales, K; and Pa ul, S. Does my mouse have Alzheimers disease? Genes, Brain, and Behavior 1: 142-155. 2002. 38. Dumont M, Strazielle C, St aufenbiel M, Lalonde R. Spa tial learning and exploration of environmental stimuli in 24-month-old female APP23 transgenic mice with the Swedish mutation. Brain Res 1024: 113-21. 2004. 39. Durlach PJ. The effects of a low dose of caffeine on cognitive performance. Psychopharmacology 140: 116-9. 1998. 40. Evenden JL, Turpin M, Oliver L, Jennings C. Caffeine and nicotine improve visual tracking by rats: a comparison with am phetamine, cocaine and apomorphine. Psychopharmacology 110: 169-76. 1993. 41. Feldman, H; Scheltens, P; S carpini, E; Hermann, N; Mesenbrink, P; Mancione, L; Tekin, S; Lane, R; and Ferris, S. Behavioral Symptoms in MCI. Neurology 62: 1199-1201. 2004. 42. Ferre S, Fredholm BB, Morelli M, Popoli P, Fuxe K. Adenosine-dopamine receptorreceptor interactions as an integrat ive mechanism in the basal ganglia. Trends Neurosci 20: 482-7. 1997. 43. Fiebich BL, Biber K, Lieb K, van Calker D, Berger M, Bauer J, Gebicke-Haerter PJ. Cyclooxygenase-2 expression in rat micr oglia is induced by adenosine A2areceptors. Glia 18: 152-60. 1996. 44. Fisone, G; Borgkvist, A; a nd Usiello, A. Caffeine as a psychomotor stimulant: mechanism of action. Cell Mol Life Sci 61: 857-872. 2004.

PAGE 130

121 45. Flagmeyer I, Haas HL, Stevens DR. Adenosine A1 receptor-mediated depression of corticostriatal and thalamostriatal glut amatergic synaptic potentials in vitro. Brain Res 778: 178-85. 1997. 46. Forbes, K; Shanks, M; and Venneri, A. The Evolution of Dysgraphia in AD. Brain Research Bull 63: 19-24. 2004. 47. Fratiglioni, L; Paillard-Borg, S; and Winblad, B. An active and socially integrated lifestyle in late life migh t protect against dementia. Lancet Neurol. 6: 343-53. 2004. 48. Frautschy, SA; Yang, F; Irrizarry, M; Hy man, B; Saido, TC; Hsaio, K; and Cole, GM. Microglial response to amyloid plaques in APPsw transgenic mice. Am J Pathol 152: 307-317. 1998. 49. Fredholm, BB; Abbracchio, MP; Burnstock, G; Daly, JW; Harden, TK; Jacobsen, KA; et al. Nomenclature and classification of purinoceptors. Pharmacol Rev 46: 143-156. 1994. 50. Fredholm, B; Battig, K; Holmen, J; Nehli g, A; and Zvartau, E. Actions of caffeine in the brain with special reference to fact ors that contribute to its widespread use. Pharmacol Rev 51: 83-133. 1999. 51. Fuso A, Seminara L, Cavallaro RA, D'Anselmi F, and Scarpa S. Sadenosylmethionine/homocyste ine cycle alterations modify DNA methylation status with consequent deregulation of PS1 and BACE and beta-amyloid production. Mol Cell Neurosci 28: 195-204. 2005. 52. Games, D; Adams, D; Alessandrini, R; Barbour, R; Berthelette, P; Blackwell, C; Carr, T; Clemens, J; Donaldson, T; G illespie, F; Guido, T; Hagopian, S; JohnsonWood, K; Khan, K; Lee, M; Leibowitz, P; Lieberburg, I; Little, S; Masliah, E; McConlogue, L; Montoya-Zavala, M; Mucke, L; Paganini, L; Penniman, E; Power, M; Schenk, D; Seubert, P; Snyder, B; Sori ano, F; Tan, H; Vitale, J; Wadsworth, S; Wolozin, B; and Zhao, J. Alzheimer-type neuropathology in transgenic mice overexpressing V717F -amyloid precursor protein. Nature 373: 523-7. 1995. 53. Gaskell, P; and Vance, J. Alzheimers Disease Genes and Genetic Testing in clinical practice. JAAPA 3: 25-31. 2004. 54. Gevaerd MS, Takahashi RN, Silveira R, Da Cunha C. Caffeine reverses the memory disruption induced by intra-nigr al MPTP-injection in rats. Brain Res Bull 55: 101-6. 2001. 55. Goncalves ML, Cunha RA, Ribeiro JA. Ade nosine A2A receptors facilitate 45Ca2+ uptake through class A calcium channels in rat hippocampal CA3 but not CA1 synaptosomes. Neurosci Lett 238: 73-7. 1997.

PAGE 131

122 56. Gordon MN, Holcomb LA, Jantzen PT, DiCa rlo G, Wilcock D, Boyett KW, Connor K, Melachrino J, O'Callaghan JP, and Morgan D. Time course of the development of Alzheimer-like pathology in the doubly transgenic PS1+APP mouse. Experimental Neuro 173: 183-195. 2002. 57. Griffiths RR, Evans SM, Heishman SJ Preston KL, Sannerud CA, Wolf B, Woodson PP. Low-dose caffeine physical dependence in humans. J Pharmacol Exp Ther 255: 1123-32. 1990. 58. Gu, JG; Foga, IO; Parkinson, FE; and Geiger, JD. Involvement of bidirectional adenosine transporters in the release of L-{H 3 }adenosine from rat brain synaptosomal preparations. J Neurochem 64: 2105-2110. 1995. 59. Halldner L, Aden U, Dahlberg V, Joha nsson B, Ledent C, Fredholm BB. The adenosine A1 receptor contributes to the stim ulatory, but not the inhibitory effect of caffeine on locomotion: a study in mice lacking adenosine A1 and/or A2A receptors. Neuropharmacology 46: 1008-17. 2004. 60. Hameleers PA, Van Boxtel MP, Hogervorst E, Riedel WJ, Houx PJ, Buntinx F, Jolles J. Habitual caffeine consumption and its relation to memory, attention, planning capacity and psychomotor perf ormance across multiple age groups. Hum Psychopharmacol 15: 573-581. 2000. 61. Hardy, J; and Selkoe, D. The Amyloid Hypothesis of Alzheimers disease: Progress and Problems on the Road to Therapeutics. Science 297: 353-356. 2002 62. Hebert, L; Scherr, P; Bienias, J; Bennett, D; and Evans, D. Alzheimer disease in the US Population. Arch Neurol 60: 1119-1122. 2003. 63. Herve D, Le Moine C, Corvol JC, Belluscio L, Ledent C, Fienberg AA, Jaber M, Studler JM, Girault JA. Galpha(olf) levels are regulated by receptor usage and control dopamine and adenosine action in the striatum. J Neurosci 21: 4390-9. 2001. 64. Higashi T, Sone Y, Ogawa K, Kitamura YT, Saiki K, Sagawa S, Yanagida T, Seiyama A. Changes in regional cerebral blood volume in frontal cortex during mental work with and without caffeine intake: functional monitoring using nearinfrared spectroscopy. J Biomed Opt 9: 788-93. 2004. 65. Hiruma, H; Katakura, T; Takahashi, S; Ichikawa, T; and Kawakami, T. Glutamate and amyloid -Protein rapidly inhibit fast axonal transport in cultured rat hippocampal neurons by different mechanisms. J Neurosci 26: 8967-77. 2003. 66. Holcomb L, Gordon MN, McGowan E, Yu X, Benkovic S, Jantzen P, Wright K, Saad I, Mueller R, Morgan D, Sanders S, Zehr C, O'Campo K, Hardy J, Prada CM,

PAGE 132

123 Eckman C, Younkin S, Hsiao K, and Duff K. Accelerated Alzheimer-type phenotype in transgenic mice carrying both mutant amyloid precursor protein and presenilin 1 transgenes. Nat Med 4: 97-100. 1998. 67. Holcomb LA, Gordon MN, Jantzen P, Hsiao K, Duff K, and Morgan D. Behavioral changes in transgenic mice expressing both amyloid precursor protein and presenilin-1 mutations: lack of a ssociation with amyloid deposits. Behav Genet 29: 177-85. 1999. 68. Holmes, C; Wilkinson, D; Dean, C; Veth anayagam, S; Olivieri, S; Langley, A; PanditaGunawardena, HD; Hogg, F; Clare, C; and Damms, J. The efficacy of donepezil in the treatment of neuropsychiat ric symptoms in Alzheimers disease. Neurology 63: 214-219. 2004. 69. Hsiao, K; Chapman, P; Nilsen, S; Eckma n, C; Harigaya, Y; Younkin, S; Yang, F; and Cole, G. Correlative memory deficits, A elevation, and amyloid plaques in transgenic mice. Science 274: 99-102. 1996. 70. Huang G Dragan M Freeman D Wilson JX Activation of catechol-Omethyltransferase in astrocytes stimulat es homocysteine synthesis and export to neurons. Glia 51:47-55. 2005. 71. Ishii, K; Sasaki, M; Yamaji, S; Sakamoto, S; Kitagaki, H; and Mori, E. Relatively preserved hippocampal glucose metabolism in mild Alzheimers disease. Dement Geriatr Cogn Disord. 6: 317-322. 1998. 72. Irizarry, M; Soriano, F; McNamara, M; Page, K; Sc henk, D; Games, D; and Hyman, B. A deposition is associated with neuropil changes, but not with overt neuronal loss in the human amyloid precursor protein V717F (P DAPP) transgenic mouse. J Neurosci 17: 70537059. 1997. 73. Izquierdo JA, Costas SM, Justel EA, Rabiller G. Effect of caffeine on the memory of the mouse. Psychopharmacology 61: 29-30. 1979. 74. Jack, CR; Petersen, RC; Xu, YC; Waring, SC; OBrien, PC; Tangalos, EG; Smith, GE; Ivnik, RJ; and Kokmen, E. Medial temporal atrophy on MRI in normal aging and very mild Alzheimers disease. Neurology 49: 786-794. 1997. 75. Jacobson KA, von Lubitz DK, Daly JW, Fr edholm BB. Adenosine receptor ligands: differences with acute versus chronic treatment. Trends Pharmacol Sci 17: 108-13. 1996. 76. James JE. Acute and chroni c effects of caffeine on performance, mood, headache, and sleep. Neuropsychobiology 38: 32-41. 1998.

PAGE 133

124 77. Jarvis MJ. Does caffeine intake enhance absolute levels of cognitive performance? Psychopharmacology 110: 45-52. 1993. 78. Jensen MT, Mottin MD, Cracchiolo JR, Le ighty RE, and Arendash GW. Lifelong immunization with human beta-amyloid (1-4 2) protects Alzheimer's transgenic mice against cognitive impairment throughout aging. Neuroscience 130: 667-684. 2005. 79. Johansson B, Ahlberg S, van der Ploeg I, Brene S, Lindefors N, Persson H, Fredholm BB. Effect of long term caffe ine treatment on A1 and A2 adenosine receptor binding and on mRNA levels in rat brain. Naunyn-Schmiedeburgs Arch Pharmacol 347: 407-414. 1993. 80. Kalback W, Watson MD, Kokjohn TA, Kuo YM, Weiss N, Luehrs DC, Lopez J, Brune D, Sisodia SS, Staufenbiel M, Emmerling M, and Roher AE. APP transgenic mice Tg2576 accumulate Abeta peptides that are distinct from the chemically modified and insoluble peptides deposited in Alzheimer's disease senile plaques. Biochemistry 41: 922-928. 2002. 81. Kamal A, Almenar-Queralt A, LeBlanc JF, Roberts EA, Goldstein LS. Kinesinmediated axonal transport of a membrane compartment containing beta-secretase and presenilin-1 requires APP. Nature 414: 643-8. 2001. 82. Kaplan GB, Greenblatt DJ, Ehrenberg BL Goddard JE, Cotreau MM, Harmatz JS, Shader RI. Dose-dependent pharmacokineti cs and psychomotor effects of caffeine in humans. J Clin Pharmacol 37: 693-703. 1997. 83. Karcz-Kubicha M, Antoniou K, Terasmaa A, Quarta D, Solinas M, Justinova Z, Pezzola A, Reggio R, Muller CE, Fuxe K, Goldberg SR, Popoli P, Ferre S. Involvement of adenosine A1 and A2A receptors in the motor effects of caffeine after its acute and chronic administration. Neuropsychopharmacology 28: 12811291. 2003. 84. Kawarabayashi, T; Younkin, LH; Saido, TC ; Shoji, M; Ashe, KH; and Younkin SG. Age-dependent changes in brai n, CSF, and plasma amyloid protein in the Tg2576 transgenic mouse model of Alzheimers disease. J Neurosci 21: 372-381. 2001. 85. Kelly PH, Bondolfi L, Hunziker D, Schlecht HP, Carver K, Maguire E, Abramowski D, Wiederhold KH, Sturchler-P ierrat C, Jucker M, Bergmann R, Staufenbiel M, and Sommer B. Progressi ve age-related impairment of cognitive behavior in APP23 transgenic mice. Neurobiol Aging 24: 365-78. 2003. 86. Kennedy, BP; Bottiglieri, T; Arning, E; Zieg ler, MG; Hansen, LA; and Masliah, E. Elevated S-adenosylhomocysteine in Alzheimer brain: influence on methyltransferases and cognitive function. J Neural Transm 111: 547-567. 2004.

PAGE 134

125 87. King, DL; Arendash, GW; Crawford, F; St erk, T; Menendez, J; and Mullan, MJ. Progressive and gender-dependent cognitive impairment in the APP sw transgenic mouse model of Alzheimers disease. Behavior Brain Res 103: 145-162. 1999. 88. King, DL; and Arendash, GW. Behavioral ch aracterization of the Tg2576 transgenic mouse model of Alzheimers disease through 19 months. Physio Beh 75: 627-642. 2002. 89. Kittner, B; Rossner, M; and Rother, M. Clinical trials in dementia with propentofylline. Ann NY Acad Sci 826: 307-16. 1997. 90. Klunk WE, Engler H, Nordberg A, Wang Y, Blomqvist G, Holt DP, Bergstrom M, Savitcheva I, Huang GF, Estrada S, Ause n B, Debnath ML, Barletta J, Price JC, Sandell J, Lopresti BJ, Wall A, Koivisto P, Antoni G, Mathis CA, and Langstrom B. Imaging brain amyloid in Alzheimers disease with Pitts burgh Compound-B. Ann Neurol. 55: 306-19. 2004. 91. Koistinaho M, Ort M, Cimadevilla JM, Vondrous R, Cordell B, Koistinaho J, Bures J, and Higgins LS. Specific spatial learning de ficits become severe with age in beta amyloid precursor protein transgenic mice that harbor diffuse beta -amyloid deposits but do not form plaques. Proc Natl Sci USA 98: 14675-14680. 2001. 92. Koshiishi C Kato A Yama S Crozier A Ashihara H A new caffeine biosynthetic pathway in tea leaves: utilisation of adenosine released from the S-adenosyl-Lmethionine cycle. FEBS Lett 499: 50-4. 2001. 93. Kruman II, Kumaravel TS, Lohani A, Pedersen WA, Cutler RG, Kruman Y, Haughey N, Lee J, Evans M, Mattson MP. Folic acid deficiency and homocysteine impair DNA repair in hippocampal neurons and sensitize them to amyloid toxicity in experimental models of Alzheimer's disease. J Neurosci 22: 1752-62. 2002. 94. Kuo YM, Kokjohn TA, Beach TG, Sue LI, Brune D, Lopez JC, Kalback WM, Abramowski D, Sturchler-Pierrat C, Staufenbiel M, and Roher AE Comparative analysis of amyloid-beta chemical st ructure and amyloid plaque morphology of transgenic mouse and Alzheimer's disease brains. J Biol Chem 276: 12991-12998. 2001. 95. Lalonde R, Dumont M, Staufe nbiel M, Sturchler-Pierrat C, Strazielle C. Spatial learning, exploration, anxiety, and motor c oordination in female APP23 transgenic mice with the Swedish mutation. Brain Res 956: 36-44. 2002. 96. Larson, J; Lynch, G; Games, D; and Seubert, P. Alterations in synaptic transmission and long-term potentiation in hippocampa l slices from young and aged PDAPP mice. Brain Research 840: 23-35. 1999.

PAGE 135

126 97. Leighty RE, Nilsson LN, Potter H, Costa DA, Low MA, Bales KR, Paul SM, Arendash GW. Use of multimetric statistical analysis to characterize and discriminate between the performance of four Alzheimer's transgenic mouse lines differing in Abeta deposition. Behav Brain Res 153: 107-21. 2004. 98. Linden J. Molecular approach to ad enosine receptors: receptor-mediated mechanisms of tissue protection. Annu Rev Pharmacol Toxicol 41: 775-87. 2001. 99. Lorenzo A, Yuan M, Zhang Z, Paganetti PA, Sturchler-Pierrat C, Staufenbiel M, Mautino J, Vigo FS, Sommer B, and Yankner BA. Amyloid beta interacts with the amyloid precursor protein: a potential t oxic mechanism in Alzheimer's disease. Nat Neurosci 3: 460-464. 2000. 100. MacDonald RL, Skerritt JH, and Werz MA. Adenosine agonists reduce voltagedependent calcium conductance of mouse sensory neurones in cell culture. J Physio 370: 75-90. 1986. 101. Maia, L, and de Mendonca, A. Does ca ffeine intake protect from Alzheimers disease? Euro J Neuro 9: 377-382. 2002. 102. Masliah, E; Sisk, A; Mallor y, M; Mucke, L; Schenk, D; and Games, D. Comparison of neurodegenerative pathology in tr ansgenic mice overexpressing V717F amyloid precursor protein and Alzheimers disease. J Neurosci 16: 57955811. 1996. 103. Mathis, CA; Wang, Y; and Klunk, WE. Im aging beta-amyloid plaques and neurofibrillary tangles in the aging human brain. Curr Pharm Des. 13:1469-92. 2004. 104. Matsuoka Y, Picciano M, Malester B, La Francois J, Zehr C, Daeschner JM, Olschowka JA, Fonseca MI, O'Banion MK, Tenner AJ, Lemere CA, Duff K. Inflammatory responses to amyloidosis in a transgenic mouse model of Alzheimer's disease. Am J Pathol 158: 1345-54. 2001. 105. Mattson, M; Cheng, B; Davis, D; Bryant K; Lieberburg, I; and Rydel, R. Amyloid peptides destabilize calcium ho meostasis and render human cortical neurons vulnerable to excitotoxicity. J Neurosci 12: 376-389. 1992. 106. McPherson, PS; Kim, YK; Valdivia, H; Knudson, CM; Takekura, H; FranziniArmstrong C; et al. The brain ryanodine receptor: a caffeine-sensitive calcium release channel. Neuron 7: 17-25. 1991. 107. Meyer L, Caston J. Stress alters caffein e action on investigatory behaviour and behavioural inhibition in the mouse. Behavioral Brain Research 149: 87-93. 2004.

PAGE 136

127 108. Miller JW Shukitt-Hale B Villalobos-Molina R Nadeau MR Selhub J Joseph JA Effect of L-Dopa and the catechol-O-m ethyltransferase inhibitor Ro 41-0960 on sulfur amino acid metabolites in rats. Clin Neuropharmacol 20: 55-66. 1997. 109. Miner, SE; Evrovski, J; and Cole, D. C linical chemistry and molecular biology of homocysteine metabolism: An update. Clin Biochem 30: 189-201. 1997. 110. Mizrahi EH, Jacobsen DW, Debanne SM, Tr aore F, Lerner AJ, Friedland RP, and Petot GJ. Plasma total homocysteine levels, dietary vitamin B6 a nd folate intake in AD and healthy aging. J Nutr Health Aging 7: 160-165. 2003. 111. Molinengo L, Scordo I, Pastorello B. Action of caffeine, L-PIA and their combination on memory retention in the rat. Life Sci 54: 1247-50. 1994. 112. Moran, M; Walsh, C; and Lynch, A. Syndrom es of behavioral and psychological symptoms in mild AD. Int J Geriatr Psychiatry 19: 359-364. 2004 113. Morgan D, Diamond DM, Gottschall PE, Ugen KE, Dickey C, Hardy J, Duff K, Jantzen P, DiCarlo G, Wilcock D, Connor K, Hatcher J, Hope C, Gordon M, Arendash GW. A beta peptide vaccination prevents memory loss in an animal model of Alzheimer's disease. Nature 408: 982-5. 2000. 114. Morris MS. Homocysteine and Alzheimer's disease. Lancet Neurol 2: 425-428. 2003. 115. Morrison LD, Smith DD, Kish SJ. Brain S-adenosylmethionine levels are severely decreased in Alzheimer's disease. J Neurochem 67: 1328-31. 1996. 116. Mullan, M; Crawford, F; Axelman, K; Houlden, H; Lilius, L; Winblad, B; and Lannfelt, L. A pathogenic mutation for pr obable Alzheimers disease in the APP Gene at the N-terminus of beta-amyloid. Nat Genet 1: 345-347. 1992. 117. Myers MG. Effect of caffeine on bl ood pressure beyond the laboratory. Hypertension 43: 724-5. 2004. 118. Nehlig A, Boyet S. Dose-response study of caffeine effects on cerebral functional activity with a specific focus on dependence. Brain Res 858: 71-7. 2000. 119. Nestor, P; Scheltens, P; and Hodges, J. Advances in the early detection of Alzheimers disease. Nature Reviews Neuroscience 5: S34-S41. 2004. 120. Nilsson, LN; Arendash, GW; Leighty, RE ; Costa, DA; Low, MA; Garcia, MF; Cracciolo, JR; Rojiani, A; Wu, X; Bales, KR; Paul, SM; and Potter, H. Cognitive Impairment in PDAPP mice depends on ApoE and ACT-catalyzed amyloid formation. Neurobiol Aging 25:1153-1167. 2004.

PAGE 137

128 121. Ohta A, Sitkovsky M. Role of G-protein-coupled adenosine receptors in downregulation of inflammation a nd protection from tissue damage. Nature 414: 916-20. 2001. 122. Okada M, Mizuno K, Kaneko S. Adenosine A1 and A2 receptors modulate extracellular dopamine le vels in rat striatum. Neurosci Lett 212: 53-56. 1996. 123. Ongini, E. and Schubert, P. Neuropr otection induced by stimulating A 1 or blocking A 2A adenosine receptors: an apparent paradox. Drug Develop Res 45: 387-393. 1998. 124. O'Regan MH, Simpson RE, Perkins LM, Phillis JW. The selective A2 adenosine receptor agonist CGS 21680 enhances excitatory transmitter amino acid release from the ischemic rat cerebral cortex. Neurosci Lett 138: 169-72. 1992. 125. Pappolla, MA; Chyan, YJ; Omar, RA.; Hsiao, K; Perry, G; Smith, MA; and Bozner, P. Evidence of oxidative stress and in vivo neurotoxicity of beta-amyloid in a transgenic mouse model of Alzheimer's di sease: a chronic oxid ative paradigm for testing antioxidant therapies in vivo. Am J Pathol 152: 871-7. 1998. 126. Paris D, Town T, Parker TA, Tan J, Humphrey J, Crawford F, and Mullan M. Inhibition of Alzheimer's beta-amyloid induced vasoactivity and proinflammatory response in microglia by a cGMP-dependent mechanism. Exp Neurol 157: 211-21. 1999. 127. Patat A, Rosenzweig P, Enslen M, Trocherie S, Miget N, Bozon MC, Allain H, Gandon JM. Effects of a new slow release formulation of caffeine on EEG, psychomotor and cognitive functions in sleep-deprived subjects. Hum Psychopharmacol 15: 153-70. 2000. 128. Petersen, R; Doody, R; Kurz, A, Mohs, R; Mo rris, J; Rabins, P; Ritchie, K; Rossor, M; Thal, L; and Winblad, B. Current concepts in MCI. Arch Neurol 58: 1985-1992. 2001. 129. Polidori, MC. Oxidative stress and risk f actors for Alzheimers disease: clues to prevention and therapy. J Alzheimers Dis. 2: 185-91. 2004. 130. Pompl PN, Mullan MJ, Bjugstad K, Arendash GW. Adaptation of the circular platform spatial memory task for mice: us e in detecting cognitive impairment in the APP(SW) transgenic mouse model for Alzheimer's disease. J Neurosci Methods 87: 87-95. 1999.

PAGE 138

129 131. Popoli P, Betto P, Reggio R, Ricciarello G. Adenosine A2A receptor stimulation enhances striatal extracellular glutamate levels in rats. Eur J Pharmacol 287: 215217. 1995. 132. Popoli P, Pintor A, Domenici MR, Frank C, Tebano MT, Pezzola A, Scarchilli L, Quarta D, Reggio R, Malchiodi-Albedi F, Falchi M, Massotti M. Blockade of striatal adenosine A2A receptor reduces, through a presynaptic mechanism, quinolinic acid-induced excitotoxicity: possible relevance to neuroprotective interventions in neurodegenerative diseases of the striatum. J Neurosci 22: 1967-75. 2002. 133. Potter, H; Wefes, IM; Nilsson, LN. The inflammation-induced pathological chaperones ACT and apo-e are necessary ca talysts of Alzheimer amyloid formation. Neurobio Aging 22: 923-30. 2001. 134. Quadri, P; Fragiacomo, C; Pezzati, R; Zanda, E; Forloni, G; Tettamanti, M; and Lucca, U. Homocysteine, folate, and vitamin B-12 in mild cognitive impairment, Alzheimers disease, and vascular dementia. Am J Clin Nutr. 80: 114-22. 2004. 135. Quadros A, Patel N, Crescentini R, Crawfo rd F, Paris D, and Mullan M. Increased TNFalpha production and Cox-2 activity in organotypic brain slice cultures from APPsw transgenic mice. Neurosci Lett 353: 66-68. 2003. 136. Rainnie DG, Grunze HC, McCarley RW, Greene RW. Adenosine inhibition of mesopontine cholinergic neurons: implications for EEG arousal. Science 263: 68992. 1994. 137. Ribeiro, JA; Sebastiao, AM; and de Mendon ca, A. Adenosine receptors in the nervous system: pathophys iological implications. Prog Neurobiol 68: 377-392. 2003. 138. Rieg T, Steigele H, Schnermann J, Richte r K, Osswald H, Vallon V. Requirement of intact adenosine A1 receptors for the diuretic and natriuretic action of the methylxanthines theophylline and caffeine. J Pharmacol Exp Ther [Epub ahead of print]: 2004. 139. Ringheim GE. Glial modulating and neurotr ophic properties of propentofylline and its application to Alzheimer's disease and vascular dementia. Ann NY Acad Sci 903: 529-34. 2000. 140. Rivkees SA, Zhao Z, Porter G, and Turner C. Influences of adenosine on the fetus and newborn. Mol Genet Metab 74: 160-171. 2001. 141. Robertson D, Hollister AS, Kincaid D, Workman R, Goldberg MR, Tung CS, Smith B. Caffeine and hypertension. Am J Med 77: 54-60. 1984.

PAGE 139

130 142. Ross, GW; Abbott, RD; Petrovitch, H; Mo rens, DM; Grandinetti, A; Tung, KH; Tanner, CM; Masaki, KH; Blanchette, PL; Curb, JD; Popper, JD; and White, LR. Association of coffee and caffeine intake with the risk of Parkinsons disease. JAMA 20: 2674-9. 2000. 143. Rudolphi KA, Keil M, Hinze HJ. Effect of theophylline on ischemically induced hippocampal damage in Mongolian gerbils: a behavioral and histopathological study. J Cereb Blood Flow Metab 7: 74-81. 1987. 144. Rudolphi KA, Keil M, Fastbom J, Fredholm BB. Ischaemic damage in gerbil hippocampus is reduced following upregulation of adenosine (A 1) receptors by caffeine treatment. Neurosci Lett 103: 275-280. 1989. 145. Sadowski M, Pankiewicz J, Scholtzova H, Ji Y, Quartermain D, Jensen CH, Duff K, Nixon RA, Gruen RJ, Wisniewski T. Amyloid-beta deposition is associated with decreased hippocampal glucose metabolism and spatial memory impairment in APP/PS1 mice. J Neuropathol Exp Neurol 63: 418-28.2004. 146. Sansone M, Battaglia M, Castellano C. Effect of caffeine and nicotine on avoidance learning in mice: lack of interaction. J Pharm Pharmacol 46: 765-767. 1994. 147. Schenk, D; Masliah, E; Lee, M; Johnson-W ood, K; Seubert, P; and Games, D. The PDAPP transgenic mouse as an animal model for A -induced amyloidosis and neuropathology. Alzheimers Disease Rev 2: 20-27. 1997. 148. Schiffmann SN, Libert F, Va ssart G, Vanderhaeghen JJ. Distribution of adenosine A2 receptor mRNA in the human brain. Neursci Lett 130: 177-81. 1991. 149. Schmitt JA, Hogervorst E, V uurman EF, Jolles J, Riedel WJ. Memory functions and focussed attention in middle-aged and el derly subjects are unaffected by a low, acute dose of caffeine. J Nutr Health Aging 7: 301-3. 2003. 150. Schubert, P; Morino, T; Miyazaki, H; Ogat a, T; Nakamura, Y; Marchini, C; and Ferroni, S. Cascading Glia Reactions : A common pathomechanism and its differentiated control by cy clic nucleotide signaling. Ann NY Acad Sci 903: 24-33. 2000. 151. Schwab, C; Hosokawa, M; and McGeer, PL. Transgenic mice overexpressing amyloid beta protein are an incomplete model of Alzheimer disease. Exp Neurology 188: 52-64. 2004. 152. Seabrook, G; and Rosahl, T. Transgenic an imals relevant to Alzheimers disease. Neuropharm 38:1-17. 1999.

PAGE 140

131 153. Sebastiao AM, Ribeiro JA. Adenosine A2 receptor-mediated excitatory actions on the nervous system. Prog Neurobiol 48: 167-89. 1996. 154. Selkoe, D. Alzheimers Disease: Genes, Proteins, and Therapy. Physiological Reviews 81: 741-766. 2001. 155. Selley, ML. Increased homocysteine a nd decreased adenosine formation in Alzheimers disease. Neurol Res 26: 554-7. 2004. 156. Seshadri S Beiser A Selhub J Jacques PF Rosenberg IH D'Agostino RB Wilson PW Wolf PA Plasma homocysteine as a risk factor for dementia and Alzheimer's disease. New Engl J Med 346: 476-483. 157. Shi D, Nikodijevic O, Jacobson KA, Daly JW Chronic caffeine alters the density of adenosine, adrenergic, cholinergic, GAB A, and serotonin receptors and calcium channels in mouse brain. Cell Mol Neurobiol 13: 247-261. 1993. 158. Shi J, Benowitz NL, Denaro CP, Sheiner LB. Pharmacokinetic-pharmacodynamic modeling of caffeine: tolerance to pressor effects. Clin Pharmacol Ther 53: 6-14. 1993b. 159. Shi D, Daly JW. Chronic effects of xanthines on levels of central receptors in mice. Cell Mol Neurobiol 19: 719-732. 1999. 160. Silva RH, Frussa-Filho R. The plus-maze discriminative avoidance task: a new model to study memory-anxi ety interactions. Effects of chlordiazepoxide and caffeine. J Neurosci Methods 102: 117-125. 2000. 161. Sobow, T; Flirski, M; and Liberski, PP. Amyoid-beta and tau proteins as biochemical markers of Alzheimers disease. Acta Neurobiol Exp 64: 53-70. 2004. 162. Sturchler-Pierrat C, Abramowski D, Duke M, Wiederhold KH, Mistl C, Rothacher S, Ledermann B, Burki K, Frey P, Paganetti PA, Waridel C, Calhoun ME, Jucker M, Probst A, Staufenbiel M, and Sommer B. Two amyloid precursor protein transgenic mouse models with Alzheimer disease-like pathology. Proc Natl Acad Sci USA 94: 13287-92. 1997. 163. Sullivan GW. Adenosine A2A receptor agonists as anti-inflammatory agents. Curr Opin Investig Drugs 4: 1313-9. 2003. 164. Svenningsson P, Le Moine C, Fisone G, Fredholm BB. Distri bution, biochemistry and function of striatal adenosine A2A receptors. Prog Neurobiol 59: 355-96. 1999. 165. Takeuchi A, Irizarry MC, Duff K, Sai do TC, Hsiao Ashe K, Hasegawa M, Mann DM, Hyman BT, and Iwatsubo T. Age-related amyloid beta deposition in transgenic

PAGE 141

132 mice overexpressing both Alzheimer mutant pr esenilin 1 and amyloid beta precursor protein Swedish mutant is not associated with global neuronal loss. Am J Pathol 157: 331-339. 2000. 166. Tanaka, Y; Sakurai, M; Goto, M; and Hayashi, S. Eff ect of xanthine derivatives on hippocampal long-term potentiation. Brain Res 522: 63-68. 1990. 167. Tanaka, H; Nakazawa, K; Arima, M; and Iwasaki, S. Caffeine and its dimethylxanthines and fetal ce rebral development in rat. Brain Dev 6: 355-361. 1984. 168. Tandon, A; and Fraser, P. The presenilins. Genome Biol 3: 3014-3020. 2002. 169. Tariot. PN; Farlow, MR; Grossberg, GT; Graham, SM; McDonald, S; and Gergel, I. Memantine treatment in patients with m oderate to severe Alzheimers disease already receiving donepezil: a randomized controlled trial. JAMA 291: 317-24. 2004. 170. Trinchese F, Liu S, Battaglia F, Walter S, Mathews PM, Arancio O. Progressive age-related development of Alzhei mer-like pathology in APP/PS1 mice. Ann Neurol 55: 801-14. 2004. 171. Trussell LO, and Jackson MB. Adenosin e-activated potassium conductance in cultured striatal neurons. Proc Natl Acad Sci 82: 4857-61. 1985. 172. Tsutsui S, Schnermann J, Noorbakhsh F, Henry S, Yong VW, Winston BW, Warren K, Power C. A1 adenosine receptor upr egulation and activation attenuates neuroinflammation and demyelination in a model of multiple sclerosis. J Neurosci 24: 1521-1529. 2004. 173. Ulas J, Brunner LC, Nguyen L, and Cotman CW. Reduced density of adenosine A1 receptors and preserved coupling of ade nosine A1 receptors to G proteins in Alzheimer hippocampus: a quantitative autoradiographic study. Neuroscience 52: 843-54. 1993. 174. van Boxtel MP, Schmitt JA, Bosma H, Jolles J. The effects of habitual caffeine use on cognitive change: a longitudinal perspective. Pharmacol Biochem Behav 75: 921-7. 2003. 175. Van Dam D, D'Hooge R, St aufenbiel M, Van Ginneken C, Van Meir F, De Deyn PP. Age-dependent cognitive decline in the APP23 model precedes amyloid deposition. Eur J Neurosci 17: 388-96. 2003. 176. Vernikos-Danellis, J and Harris III, CG. The e ffect of in vitro and in vivo caffeine, theophylline, and hydrocortisone on the phosphodi esterase activity of the pituitary,

PAGE 142

median eminence, heart, and cerebral cortex. Proc Soc Exp Biol Med 128: 1016-1021. 1968. 177. Veurink, G; Fuller, S; Atwood, C; and Martins, R. Genetics, lifestyle, and the roles of amyloid and oxidative stress in Alzheimers disease. Ann Human Biol 30: 639-667. 2003. 178. Vinters, H. V. Stroke 18: 311 324. 1987 179. Von Lubitz DK, Paul IA, Bartus RT, Jacobson KA. Effects of chronic administration of adenosine A1 receptor agonist and antagonist on spatial learning and memory. Eur J Pharmacol 249: 271-80. 1993. 180. Wahlund, L; and Blennow, K. Cerebrospinal fluid biomarkers for disease stage and intensity in cognitively impaired patients. Neurosci Lett 339: 99-102. 2003. 181. Wallman-Johansson, A; and Fredholm, BB. Release of adenosine and other purines from hippocampal slices stimulated electrically or by hypoxia/hypoglycemia. Effect of chlormethiazole. Life Sci 55: 721-8. 1994. 182. Westerman, MA; Cooper-Blacketer, D; Mariash, A; Kotilinek, L; Kawarabayashi, T; Younkin, LH; Carlson, GA; Younkin, SG; and Ashe, KH. The relationship between A and memory in the Tg2576 mouse model of Alzheimers disease. J Neurosci 22: 1858-1867. 2002. 183. Xu, Y; Jack, CR; OBrien, PC; Kokmen, E; Smith, GE; Ivnik, RJ; Boeve, BF; Tangalos, RG; and Petersen, RC. Usefulness of MRI measures of entorhinal cortex versus hippocampus in AD. Neurology 54: 1760-1767. 2000. 184. Yeomans MR, Ripley T, Davies LH, Rusted JM, Rogers PJ. Effects of caffeine on performance and mood depend on the level of caffeine abstinence. Psychopharmacology 164: 241-9. 2002. 185. Zekanowski, C; Religa, D; Graff, C; Filipek, S; and Kuznicki, J. Genetic aspects of Alzheimers disease. Acta Neurobiol Exp 64: 19-31. 2004 186. Zimmermann, H; and Braun, N. Ecto-nucleotides-molecular structures, catalytic properties, and functional roles in the nervous system. Prog Brain Res 120: 371-385. 1999. 133

xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001912709
003 fts
005 20071011100216.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 071011s2005 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001342
QH307.2 (ONLINE)
1 100
Schleif, William.
0 245
Effects of long-term administration of caffeine in a mouse model for Alzheimer's disease
h [electronic resource] /
by William Schleif.
[Tampa, Fla] :
b University of South Florida,
3 520
ABSTRACT: A recent epidemiological study suggested that higher caffeine intake reduces the risk of Alzheimer's disease (AD). Caffeine, a widely consumed stimulatory drug, is a non-selective adenosine receptor antagonist that has been shown to increase plasma adenosine levels in rodents. To determine any long-term protective effects of caffeine in a controlled longitudinal study, caffeine was added to the drinking water of APPsw transgenic (Tg) mice between 4 and 9 1/2 months of age, with behavioral testing done during the last 6 weeks of treatment. The average daily intake of caffeine per mouse (1.5 mg) was the human equivalent of 5 cups of coffee/day. Across multiple cognitive tasks of spatial learning/reference memory, working memory, and recognition/identification, Tg mice given caffeine (Tg+Caff) performed significantly better than Tg control mice and similar to non-transgenic controls. Discriminant Function Analysis involving multiple cognitive measures clearly showed the^ superior overall cognitive performance of Tg+Caff mice compared to Tg controls. Analysis of amyloid beta in the hippocampus by ELISA revealed Tg+Caff mice had significantly less soluble amyloid beta 1-40 and insoluble amyloid beta 1-42. In a follow-up study involving neurochemical analysis only, caffeine was added to the drinking water of 17 month old APPsw mice for 18 days. In this study, Tg+Caff mice also showed a significant reduction of insoluble amyloid beta 1-42 in the hippocampus. In contrast to the reduced extracellular brain levels of adenosine in Tg controls, caffeine treatment normalized brain adenosine levels in Tg mice to that of non-transgenic controls. Analysis of amyloidogenic secretase activity revealed the reduction in amyloid beta is likely because of a reduction in gamma secretase activity as a result of increased SAM silencing of PS1 expression. This study suggest that a modest, long-term caffeine intake of approximately 500 mg per day (5 cups of coffee) may redu ce considerably the risk of AD by decreasing amyloidogenesis.
Thesis (M.A.)--University of South Florida, 2005.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 133 pages.
Adviser: Gary Arendash, Ph.D.
Transgenic mice.
Dissertations, Academic
x Biology
t USF Electronic Theses and Dissertations.
4 856