Evidence is accumulating that chronic inflammation exacerbates the pathology in a number of degenerative diseases of aging including Parkinson's disease (PD), diffuse Lewy body disease (DLB), Alzheimer's disease (AD), progressive supranuclear palsy (PSP), amyotrophic lateral sclerosis (ALS), multiple system atrophy (MSA), the fronto-temporal dementias, type II diabetes, and age-related macular degeneration. Many elements are involved in these inflammatory reactions and there is now a massive literature on the subject. This review will not cover all these aspects but will concentrate on reactions of microglia, astrocytes, and oligodendrocytes in PD and related extrapyramidal disorders.
Dopaminergic neurons of the substantia nigra are particularly vulnerable to oxidative and inflammatory attack. Such processes may play a crucial role in the etiology of Parkinson disease (PD). Since glia are the main generators of these processes, the possibility that PD may be caused by glial dysfunction needs to be considered. This review concentrates on glial reactions in PD. Reactive astrocytes and reactive microglia are abundant in the substantia nigra (SN) of PD cases indicating a robust inflammatory state. Glia normally serve neuroprotective roles but, given adverse stimulation, they may contribute to damaging chronic inflammation. Microglia, the phagocytes of brain, may be the main contributors since they can produce large numbers of superoxide anions and other neurotoxins. Their toxicity towards dopaminergic neurons has been demonstrated in tissue culture and various animal models of PD. The MPTP and α-synuclein models are of particular interest. Years after exposure to MPTP, inflammation has been observed in the SN. This has established that an acute insult to the SN can result in a sustained local inflammation. The α-synuclein model indicates that an endogenous protein can induce inflammation, and, when overexpressed, can lead to autosomal dominant PD. Less is known about the role of astrocytes than microglia, but they are known to secrete both inflammatory and anti-inflammatory molecules and may play a role in modulating microglial activity. Oligodendrocytes do not seem to play a role in promoting inflammation although, like neurons, they may be damaged by inflammatory processes. Further research concerning glial reactions in PD may lead to disease-modifying therapeutic approaches. © 2007 Movement Disorder Society
Del Rio Hortega initially identified microglia in 1919 as a mesodermal cell type entering the brain in late embryonic life.1 By inflicting stab wounds, he demonstrated their reactive nature and phagocytic capacity. This identification of mesodermal cells, as well as their associated phagocytic function, was questioned for almost six decades, until immunohistochemical investigations of AD, PD, and other neurological diseases demonstrated beyond doubt that Hortega was correct.2, 3 The probable role of microglia in the pathology of PD has been the subject of a number of recent reviews.4–12
Microglia constitute about 10% of all glial cells. They are evenly distributed throughout the normal brain with their ramified processes being close together but not touching each other. This suggests that each cell takes up its own exclusive territory to patrol. Their morphology was originally described by Hortega as resting. But the recent in vivo movies of Nimmerjahn et al.13 cast a new light on microglia by demonstrating that they never rest but are in constant activity. They extend and retract their ramifications as they continuously sample the extracellular environment in their surround. When Nimmerjahn et al. injured a capillary with a laser beam, as in a miniature recapitulation of Hortega's stab wounds, the surrounding microglia responded immediately. They became activated as their processes thickened and retracted. Their surface characteristics underwent a dramatic transformation as they migrated to the site of injury and agglomerated around the injured vessel. Then, over minutes to hours, they phagocytosed the leaking blood (see supplementary information in Ref.13).
Post mortem brain specimens in PD and other degenerative diseases need to be interpreted in terms of the Nimmerjahn movies. The observed microglia reflect only their state in the terminal hours of the patient's life. What their activity was in previous days, months, or years remains completely unknown. The situation is somewhat different for neurons. The intensifying clinical symptoms of disease progression reflect accumulating neuronal deficits. Since neurons are post mitotic cells, the observed neuronal damage seen in post mortem specimens is an integral effect as contrasted with the differential effect of microglia.
Figure 1A shows a field of so-called resting microglia in normal white matter of a case of PD. In this case the microglia are revealed by an antibody to CD11b or Mac-1. CD11b is a major receptor for targets opsonized by complement. It is constitutively expressed at high levels in microglia, consistent with their role as phagocytes. Figure 1B illustrates activated microglia agglomerating around neurons and extracellular melanin in the substantia nigra (SN) of a PD patient. In this case the microglia are marked by an antibody to HLA-DR. HLA-DR is an antigen presenting glycoprotein and a marker of immunocompetent cells. It is constitutively expressed by microglia and is highly upregulated when they become activated. Figure 1C again shows activated microglia in the SN of a PD patient with the marker being an antibody to leukocyte common antigen (LCA). This demonstrates the monocytic origin of the cells. LCA is more strongly expressed on circulating leukocytes such as T-cells than on microglia. This provides an opportunity for making a cellular distinction between resident phagocytes and patrolling leukocytes in post mortem brain specimens. The figure illustrates an extracellular leukocyte near an activated microglial cell. While the numbers of such leukocytes are mildly upregulated in the SN of PD cases,14 the levels are far below what is observed in presumed autoimmune disorders such as multiple sclerosis. PD is clearly not an autoimmune disease, but evidence of localized attack by microglia places it in the autotoxic category.9
This evidence of microglial attack in PD is supported by results from three differing types of research: epidemiological, animal model, and cell culture. Epidemiological studies that investigate the effects of using anti-inflammatory agents are, in theory, the most significant. If the inflammatory reaction seen in pathological specimens is combating the disease process, individuals who consume anti-inflammatory agents should be more prone to the disease. If it is an epiphenomenon engaged only in removing debris, there should be no effect of anti-inflammatory agents. If, however, the microglial cells are attacking viable host tissue, individuals consuming anti-inflammatory agents should have a sparing of the disease.
Reports to date support the microglial attack interpretation. Chen et al.15 reported in a population based study of more than 44,000 men and 99,000 women that regular NSAID users had a 55% sparing of PD. In a follow-up study,16 they found that ibuprofen was associated with a 35% lower risk, but they found no protection from the use of other NSAIDs. Case control studies have given marginal results but, overall, are consistent with a weak protective effect of NSAID use (reviewed by Esposito et al.11).
There have been no clinical trials of NSAIDs in PD itself. Experience with AD may be instructive in this regard. The epidemiological evidence of microglial attack is more robust than in PD (reviewed by McGeer and McGeer17) but, except for promising results in pilot studies of indomethacin and diclofenac, clinical studies have been negative. However these trials have concentrated on selective COX-2 inhibitors, for which there is no supporting epidemiological data, and relatively low dose naproxen, which may be one of the ineffective NSAIDs.17
Animal model and in vitro culture studies are indirect but they do illustrate that dopaminergic cells are highly sensitive to inflammatory attack and that microglial cells can be activated to mount such an attack. Animal models of PD are generally of two types. Type 1 is based on administration of oxidizing compounds that are preferentially taken up by dopaminergic cells. Rotenone and 6-hydroxydopamine are examples. Type 2 is based on localized administration of inflammatory agents. Lipopolysaccharide (LPS) into the SN is an example.18 It is noteworthy that equivalent injections into the hippocampus and cortex did not produce local lesions, indicating the particular vulnerability of SN dopaminergic neurons. Two models are of particular interest: the MPTP and α-synuclein models.
The MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) model was discovered in a bizarre fashion. It was a minor contaminant of a street drug briefly sold to a group of heroin addicts in the San Francisco Bay area. A cluster of young users suddenly turned up with parkinsonian symptoms.19 The contaminant was soon identified as MPTP and its ability to produce parkinsonian symptoms confirmed in monkeys. The toxic material itself turned out to be 1-methyl-4-phenylpyridinium ion, a monoamine oxidase B conversion product of MPTP and a substrate for the dopamine transporter. Only a few humans were ever exposed, but all developed a fatally progressive parkinsonian disorder. Autopsies done as long as 20 years later demonstrated activated microglia in the SN similar to that observed in PD cases20 and as illustrated in Figure 1B,C.
The ability of parenterally administered MPTP to produce a sustained inflammatory reaction in the SN was confirmed in monkeys.21, 22 This is illustrated in Figure 1D, which shows activated microglia in the SN of an 18-year-old monkey that had been administered MPTP 11 years previously. Neither humans nor animals develop Lewy bodies from MPTP administration, so it is not an accurate reflection of PD. Nevertheless, the model illustrates that an acute insult to the SN can result in a sustained inflammatory response. It is therefore conceivable that in PD, as in monkeys and humans exposed to MPTP, an acute insult initiates an inflammatory reaction that becomes self sustaining after the initiating agent has disappeared.
One possible method of sustaining inflammation is through activated astrocytes expressing inflammatory mediators. An example is ICAM-1, which is a ligand for LFA-1 in activated microglia. This inflammatory combination has been shown to exist in the SN of PD cases and MPTP-treated monkeys,14 as illustrated in Figure 1E,F.
The hallmark of PD is the presence of Lewy bodies in the SN and, in the diffuse Lewy body variant of PD, in a more widespread distribution. The major component of Lewy bodies is α-synuclein.23 Three mutations in α-synuclein have been found to cause autosomal dominant PD.24 Of much greater significance is the fact that overexpression of normal α-synuclein also causes autosomal dominant PD.25 Mutations in which 3 copies of normal α-synuclein are expressed have typical onset in the fifties,26 while those in which 4 copies are expressed have typical onset in the thirties.27 This places α-synuclein in the forefront of potential causes of PD, with much attention being focused on its malfunctioning.
The discovery that α-synuclein is an inflammatory stimulant for microglia, and that animal models of PD could be based on α-synuclein exposure, is of high interest in this regard. Dopaminergic neurons as well as microglia can be harvested from embryonic and new born rodents. When activated in vitro, these microglia have been shown to contribute to the death of dopaminergic cells.11, 28–37 Zhang et al.38 reported that α-synuclein, when incubated for several days to promote oligomer formation, activated microglial cells which, in turn, were toxic towards dopaminergic neurons cultured from the brains of embryonic day 12 mice. The toxicity was reduced in mice null for NADPH oxidase (PHOX), indicating that oxygen free radicals, generated by the activated microglia, played an important role in the neuronal toxicity. They further reported that the PHOX induction was linked to direct activation of the Mac-1 receptor rather than being due to α-synuclein internalization via scavenger.39 They proposed that nigral neuronal damage, regardless of its etiology, might release aggregated α-synuclein, which could then lead to persistent and progressive neuronal damage.
Other explanations have been put forward as to why α-synuclein overexpression should induce PD,40 but this model, taken together with the MPTP model, suggests how a sustained microglial inflammation could be responsible for PD. A somewhat different conclusion was reached by Shavali et al.37 They stimulated mouse macrophage cells with LPS and found they killed SH-SY5Y cells. This was attributed to increased peroxynitrite production and nitration of α-synuclein within SH-SY5Y cells. It was suggested that death of dopaminergic neurons in PD could be caused by intraneuronal nitration of α-synuclein.
Human microglial cells can be harvested for several hours post mortem so that they can be studied in vitro. When activated by such inflammatory stimulants as interferon-γ (IFNγ) and lipopolysaccharide (LPS), their separated supernatants can be lethal to cultured human neuronal lines such as SH-SY5Y and NT2 cells.41 Such studies are important, not just for establishing that activated microglia can secrete neurotoxic substances in vivo, but also for indicating what factors might be responsible. A profusion of potential activators and neurotoxic substances have so far been identified. Table 1 lists a few of the more prominent, and presumably more powerful, of these molecules. More complete lists can be found in other reviews.47
|Material||System and reference|
|Intercellular adhesion molecule-1 (ICAM-1)||Human PD SN42, 43|
|Tumor necrosis factor α (TNF-α)||Human PD SN,42, 43 tissue culture44|
|Interleukin-6 (IL-6)||Human PD SN,28, 44 tissue culture41, 44|
|Interleukin-1β (IL-1β)||Tissue culture28|
|IL-1α and IL-1β mRNAs||SN on 6OHDA-treated rats45|
|Lymphocyte function associated antigen (LFA-1)||Human PD SN14, 42, 43|
|Brain-derived-neurotrophic factor (BDNF)||Human PD SN46|
Activated microglia have even been demonstrated in the basal ganglia and brainstem of PD cases using positron emission tomography (PET) with [11C](R)-PK11195, a marker of peripheral benzodiazepine binding sites that is selectively expressed by activated microglia.48, 49 Ouchi et al.49 did a PET analysis by comparing midbrain PK11195 binding with basal ganglia binding of the dopamine transporter marker (11C)CFT. They found a correlation in their PD cases between elevated PK11195 binding, indicative of activated microglia, and reduced CFT binding, indicative of dopaminergic terminal loss. They concluded that neuroinflammatory responses from intrinsic microglia were contributing significantly to the progressive dopaminergic degeneration in PD.
Activated microglia may produce large amounts of superoxide radicals, which may well be the major source of the oxidative stress believed to be largely responsible for dopaminergic cell death in PD.50, 51 Such oxidative stress may, for example, cause the oxidation of dopamine to the quinone products, which are believed to be damaging to brain mitochondria.52
Reactive microglia are also seen in the basal ganglia in the 6-hydroxydopamine,53, 54 MPTP,21, 55 and rotenone56, 57 animal models of PD, as well as in models of dopaminergic cell loss induced in rodents by nigral injection of LPS,58, 59 of Fcγ receptor activators such as trisialoganlioside,60 or of immunoglobulins from PD patients.11, 61 Other treatments of rodents shown to induce such activated microglia and dopaminergic cell death include systemic injection of LPS plus tumor necrosis factor-α.62
Various compounds have been shown to decrease both microglial activation and dopaminergic cell death in such animal models. Such agents include 3-hydroxymorphinan63 and minocycline.64 Mixed microglial-neuronal cultures have also been used to demonstrate the neuroprotective effects of a number of agents (Table 2). Animal models and culture studies have implicated tumor necrosis factor-α and interferon-γ, which is upregulated in PD,72 as key factors in microglial activation.73
|Agent||System||Presumed activity and reference|
|Minocycline||Rotenone-treated mouse cells||Microglial inhibitor64|
|LR1 antibody (synthetic peptide, YIGSR)a||MPP+ treated mouse cells||Microglial LR inhibition65|
|Dextromethorphan||LPS-treated mouse cells||Microglial inhibitor66|
|SB203580||Thrombin-treated rat cells||p38-MAPK2 inhibitor28b|
|PD98059||Thrombin-treated rat cells||ERK1/23 inhibitor28|
|Caffeic acid phenethyl ester||6OHDA-treated cells||Anti-inflammatory/antioxidant67|
|HU-210, a cannabinoid agonist||6OHDA-treated mouse cells||Anti-inflammatory/antioxidant68|
|Compound Ac||LPS-treated mouse cells||Anti-inflammatory69|
|XENP345||6OHDA-treated rats||TNF-α inhibitor70|
|3-Hydroxymorphinan||LPS-treated rat cells||Microglial inhibitor71d|
The evidence for a critical role of microglial NADPH oxidase in dopaminergic cell death induced by rotenone10, 56, 64, 74 is further support for the role of superoxide radicals produced by microglia. Other glial components suggested as essential for microglial activation and dopaminergic cell death are the receptors for tumor necrosis factor-α75 and the FcγR.61 A number of reports on rodent models have implicated microglial NO in cell death,58, 60, 76 but this may not apply to PD since human, unlike rodent, microglia appear to produce very little NO.77
Other microglial markers said to be involved in neurotoxicity include a deficiency in the microglial fractalkine receptor (CX3CR1),78 and an upregulation of the microglial laminin receptor.79 These receptors are suggested as new targets for anti-PD drugs.
Microglia can be activated by CC chemokine ligand 2 (CCL2, monocyte chemoattractant protein-1), which is expressed by dopaminergic neurons and can interact with its receptor CCR2 on microglial cells80; by CD40 ligand, which interacts with microglial CD4081, 82; by products of the classical complement cascade; by C-reactive protein, which is upregulated in PD; by various inflammatory cytokines; by chromogranin A,83 which has been reported to occur in PD SN84; by various forms of matrix metalloproteinase-3, which is induced in stressed dopaminergic cells35; and by aggregated α-synuclein,38 which has also been shown to be secreted by stressed neurons.85 The mutated, disease causing forms of α-synuclein are more potent stimuli of microglial activation than the wild-type protein, indicating a possible molecular mechanism for the increased toxicity of the α-synuclein mutations.85 There are reports44, 86 that human melanin can activate microglia in vitro, possibly through activation of the NF-κB receptor on such microglia,10 but injections of human melanin into rat SN failed to produce such activation.88
One report88 indicates that the extent of microglial activation, as indicated by HLA-DR staining, correlates with the amount of α-synuclein deposited in the SN in PD, but neither correlates with clinical progression. The authors use this lack of correlation to argue against inflammation as a major factor in dopaminergic cell death but the Nimmerjahn movies indicate the difficulties of long-term interpretation of microglial activity from the short term evidence of post mortem specimens.
Classical activation, which can be initiated by interferon-γ and other factors such as LPS, results in a flooding of surrounding tissue with inflammatory mediators, oxidizing free radicals, proapoptotic factors, and matrix-degrading proteases. There is emerging evidence that macrophages may also be activated to an opposite anti-inflammatory state.73 Developing reliable methods for stimulating such a state could be beneficial.
Alternative activation may be initiated by administering interleukin-4 (IL-4) or glucocorticoids. The macrophage/microglia then secrete leukocyte-attracting chemokines, anti-inflammatory cytokines, and some extracellular matrix (ECM) components (Table 3). Factors secreted by alternatively activated macrophages/microglia promote cell proliferation, angiogenesis, and ECM reconstruction, thus promoting wound repair. Anti-inflammatory type activation may be distinguished by the upregulation of CD163. CD163 (M130) is a member of the scavenger receptor cysteine-rich (SRCR) superfamily, which is expressed by monocytes and macrophages. CD163 expression is suppressed by proinflammatory mediators like LPS, interferon-γ, and tumor necrosis factor-α, whereas IL-6 and the anti-inflammatory cytokine IL-10 strongly upregulate CD163 mRNA in monocytes and macrophages.73
|Activating signals||IFN-γ, TNF||IL-4, glucocorticoids|
|Secretory products||TNFα, IL-12, IL-1, IL-6||IL-1RA, IL-10|
|Biological markers||MHC class II, CD86||CD163, mannose receptor, scavenger receptor, CD23|
|Killer molecules||NO, O2−, others||None|
|Chemokine production||IL-10, MIP-1α MCP-1||AMAC-1|
A third type of activation has been reported to be initiated by ligation of Fcγ receptors (Fcγs), coupled with a macrophage stimulatory signal through any of the Toll-like receptors (TLRs), CD40, or CD44. This results in production of large amounts of IL-10. This can even occur in IFN-γ-primed macrophages, where the IL-12 production is cut off and IL-10 induced. However, the production of many other cytokines such as TNF, IL-1, and IL-6 is unaffected.73
Much of the work on alternate forms of activation has been done on murine cells, which are easily followed because of their abundant production of NO following classical activation. It remains to be established whether similar alternative pathways can be induced in human macrophages/microglia.
Many years ago, Pergolide, a dopamine agonist effective in the treatment of PD, and clonidine, an alpha adrenergic agonist, were shown to have anti-inflammatory activity in the carrageenan paw edema assay.89 Later it was shown that microglia express a range of neurotransmitter receptors including and D1 and D2 dopaminergic receptors90 and α1A, α2A, β1, and β2 noradrenergic receptors.91, 92 Stimulation of these receptors in vitro attenuated the LPS-induced release of TNF-α and IL-6.41, 92 The possibility exists that a loss of dopaminergic neurons results as well in a loss of the protective effect of endogenous anti-inflammatory activity.
Clearly it would be beneficial in PD and the other degenerative diseases characterized by chronic inflammation to switch from proinflammatory type macrophages to anti-inflammatory type. Whether this can be done while maintaining, or even stimulating phagocytosis, remains to be established.
The functions of astrocytes are even less well understood than those of microglia. They migrate to a site of injury and develop a hypertrophic morphology. They are then described as reactive astrocytes. They appear not to attack a pathological target, as do microglia, but to wall it off. This is particularly evident in AD, where microglia agglomerate on the amyloid deposit, while astrocytes form a surrounding shell, as if to wall off the area being attacked (Fig. 1G). Their role in such an anatomical arrangement is unclear. They might be stimulating the microglia while at the same time secreting protective agents to the peripheral surround. They are known to elaborate both proinflammatory and anti-inflammatory agents.
For example, astrocytes produce factors that may be important in the inflammatory reaction that occurs in the SN in PD. Many ICAM-1 positive astrocytes are seen in the SN in PD14 (Fig. 1F) and this may attract reactive microglia to the area since such microglia carry the counter receptor LFA-1.14 α-Synuclein is capable both of activating microglia and stimulating astrocytes to produce IL-6 and ICAM-1.85 This combination can attract further microglia to the site. The action of α-synuclein on astrocytes is believed to be through receptors but the identity of these receptors is currently unknown. However antagonists of such putative α-synuclein receptors, as well as antagonists of CD40L, CCL2 and possibly other chemokines,93 might constitute novel PD-specific anti-inflammatory agents.
Astrocytes, on the other hand, have been shown to secrete a number of neurotrophic factors for dopaminergic neurons. These include glial cell-line-derived neurotrophic factor (GDNF),94–96 brain-derived neurotrophic factor (BDNF),46, 94 and mesencephalic astrocyte-derived neurotrophic factor (MANF).97 Interestingly, valproate94 and 3-hydroxymorphinan,63, 71 both of which have been reported to protect midbrain dopamine neurons against LPS or MPP+-induced neurotoxicity, are said to upregulate the production of neurotrophic factors by astrocytes as well as reduce reactive microgliosis. Astrocytes in the SN are also reported to upregulate protease-activated receptor-1 (PAR-1) in PD and this supposedly has a protective effect by increasing the activity of glutathione peroxidase.98
A very recent report indicates that astrocytes express NF-E2-related factor (Nrf2), which binds to the antioxidant response element to induce antioxidant enzymes. Overexpression of Nrf2 in astrocytes is reported to protect against 6-OHDA damage in mice,99 suggesting another approach to therapeutics aimed at reducing cell death in PD.
Astrocytic activation is also believed to play a role in the neuroprotective action of (R)-(−)-2-propyloctanoic acid in the MPTP model of PD,100 but the factors involved have not been identified.
The PARK7 (DJ-1) gene, which has been implicated in some forms of early-onset, autosomal recessive PD, is apparently expressed mainly by astrocytes in human brain.101 How this relates to the pathophysiology of PD is, however, unclear.
Astrocytes, as well as microglia, express a variety of neurotransmitter receptors.102 It has been reported that astrocytes in multiple sclerosis lack β-2 adrenergic receptors,103 but there are no reports of investigation in PD.
Astrocytes may therefore also play dual roles in PD even though there is one report104 of surprisingly low reactive astrocytosis in the SN in PD.
There is very little literature on oligodendrocytes in PD. One group did report the presence of complement-activated oligodendrocytes in the SN in PD cases105, 106 (Fig. 1H) and in many brain regions in two cases of Lewy body dementia with greatly reduced dopamine levels.105, 107 The complement-activated oligodendrocytes were revealed by immunohistochemical staining with antibodies to C3d and C4d.
Oligodendroglial, astrocytic, microglia, and neuroblastroma cell lines all have been reported to express the mRNA for the leucine-rich repeat kinase 2 gene (LRRK2),108 mutations in which have been identified in some familial PD cases. Immunostaining for the protein identifies Lewy bodies in PD and oligodendroglial coiled bodies in the Parkinson dementia complex of Guam and the N279K tau mutation, which causes pallido-ponto-nigral degeneration (PPND).108 In tauopathies, oligodendroglia may develop coiled bodies and astrocytes may become tufted as shown for tau immunostaining in a case of PPND (Fig. 1I). In both Lewy body dementia and PD, α-synuclein-containing inclusions have been reported in oligodendroglia109, 110 as well as in astrocytes.110
Dopaminergic cells of the SN are highly vulnerable to oxidative and inflammatory attack. A robust inflammatory reaction is observed in the SN in PD and related extrapyramidal disorders. Reactive microglia and reactive astrocytes are prominent in affected areas of the SN. Epidemiological studies indicate that chronic users of NSAIDs have a relative sparing of PD, supporting the concept that inflammatory attack is contributing to dopaminergic neuronal loss. Animal models of PD are based on oxidative stress or inflammatory stimulation to the SN area. The MPTP model indicates that inflammation in the SN can be self-sustaining while the α-synuclein model indicates that overexpression of this endogenous protein can be an inflammatory source. Microglia and astrocytes can secrete inflammatory cytokines and other neurotoxic products. They can also secrete neuroprotective products. Blocking inflammation or shifting the balance between proinflammatory and anti-inflammatory states offers the best current strategy for developing disease-modifying therapies for PD and related disorders.
This work was supported by a grant from the Pacific Alzheimer Research Foundation.