Alzheimer disease (AD) is believed to result from an abnormal processing of the amyloid precursor protein (APP) that leads to the accumulation of "toxic products" (TPs), most notably β-amyloid (1). There is, however, a growing recognition that the immune system may modulate APP processing, TP clearance, the cellular responses to TPs, and, therefore, that the immune system may serve as a valuable entry point for treatment (1—5). Nevertheless, considerable uncertainty remains about the exact role of the immune system in AD (2, 3). Most efforts to understand the role of the immune system and vaccines that target β-amyloid for AD treatment have focused on inflammation and the role of activated microglia in secreting cytokines and in the phagocytosis and clearing of the TPs of APP processing. More recently, however, it has become apparent that T cells, another cellular component of the immune system, found in increased numbers in postmortem brains of patients with AD (6, 7) may play an important role in regulating microglia activation and function (8).
Central to our understanding of AD and its treatment are the pathological features of neuritic plaques (NPs) and neurofibrillary tangles (NFTs) (9), the most important components of which are β-amyloid (1, 10—11) (Aβ1-40 and Aβ1-42 or Aβ) and hyperphosphorylated tau (12—15), respectively. For some years the field was split between those who believed that β-amyloid (amyloid hypothesis) was the trigger that was responsible for the neurodegeneration and those who believed that the trigger was abnormal tau (tau hypothesis). The findings that genetically dominant early-onset forms of AD resulted from mutations in APP, presenilin 1, and presenilin 2 substantially altered the debate. Mutations in all three genes could be shown to lead to variations in the processing of APP, resulting in increased concentrations of Aβ1-40 and Aβ1-42 and/or a shift in their ratio favoring the formation of Aβ1-42. Aβ1-42 was demonstrated to have greater toxic potential because of its greater propensity to form oligomers and fibrilized forms (16). However, it is yet to be determined which of the various combination of products and aggregates of the APP processing pathway as well as their location, i.e., intracellular or extracellular, are the chemical moieties and pathways that most directly affect neuronal damage in AD (17—21).
DOES THE IMMUNE SYSTEM CONTRIBUTE TO AD PATHOLOGY?
There is increasing evidence that the mere accumulation of β-amyloid (Aβ1-40 and Aβ1-42 and all their forms including diffuse amyloid plaques and NPs) may not be sufficient to induce the cognitive findings of AD and that the immune system may play a critical role in the clinical symptoms. Activated microglia and astrocytes, cellular components of the brain's immune network, are found in close proximity to senile plaques (22—25). Elevated levels of both pro- and anti-inflammatory cytokines, e.g., interleukin (IL)-1β, IL-6, tumor necrosis factor (TNF)-α, and C-reactive protein are found in the plasma and/or cerebrospinal fluid of patients with AD (26—29). Evidence of complement activation is also found in the AD brain (30) together with increased numbers of brain T cells, primarily of the CD8+ type (6, 7).
Despite the fact that the above findings were reported some time ago, the absence of a leukocyte infiltration diminished interest in the role of the immune system and neuroinflammation in AD. It was not until findings from several epidemiological studies suggested that anti-inflammatory drugs, particularly nonsteroidal anti-inflammatory drugs, reduce the risk of developing AD that interest developed in the possible causative role of neuroinflammation in the clinical symptoms of AD (31—34). Subsequent support for the importance of the immune system in AD pathology derives from the findings that polymorphisms in genes that are part of inflammatory pathways are associated with altered risk of AD (35—41) and from neuroimaging studies using labeled PK1195 (1-[2-chlorphenyl]-N-methyl-N-[1-methyl-propyl-]-3-isoquinoline carboxamide) (42, 43). PK1195 is a tracer that binds to the peripheral benzodiazepine receptor, a receptor found on activated microglia cells, cells normally present in very small numbers in the brain. An increased accumulation of the tracer has been found in entorhinal and temporoparietal cortices and the posterior cingulate of even mildly affected patients with AD. Unfortunately, however, trials of anti-inflammatory agents in AD have not shown efficacy (33, 34).
If the toxic products of APP processing are what results in AD, what are the possible mechanisms for immune system modification of clinical symptoms? Postmortem findings have suggested that clinical manifestations of AD are most closely tied to regional accumulation of NFTs rather than NPs and loss of synapses even more than NFTs (44—48). Studies of neuronal damage reveal that inflammation can modify or accelerate APP production and tau phosphorylation (49—54). Further, individuals known to be healthy before their death can have substantial amounts of β-amyloid found at postmortem (55—61), but these brains usually do not display the typical neuroinflammatory markers of AD (62). In addition, subjects with possible as well as early AD at postmortem show substantial numbers of activated microglia (63—65), suggesting that neuroinflammation can occur early in the disease process.
Further, Toll-like receptors (TLRs) have been linked to the pathogenesis of AD and TLRs 1, 2, 4, 5, 7, and 9 and CD14 are up-regulated in the aging mouse brain (66, 67). Both TLRs and CD14 are membrane proteins found on cells that are part of the immune system. In this context, it is of some significance to note that the immune system consists of two networks, the innate and the adaptive. Initiation of the innate network response occurs through the recognition of "pathogen-associated molecular patterns" that are produced by microorganisms. In general, recognition relies on interaction with TLRs and inflammation on downstream signaling to activate the transcriptional factor nuclear factor-κB that leads to increased transcription of proinflammatory genes.
Recent studies with positron emission tomography using amyloid tracers have shown that some 10%—15% of older healthy individuals have amyloid tracer accumulations that overlap with those of patients with AD and that even a higher percentage was seen in those with mild cognitive impairment (68—72). Thus, the amyloid tracer data are consistent with other neuroimaging and cognitive data, suggesting that changes in brain physiology in individuals at risk for AD may begin decades earlier than the expected clinical manifestation of the disorder (73—81) and that β-amyloid accumulation is not always immediately toxic.
Although abnormal amyloid processing is central to AD pathology, additional factors must contribute to the onset and progression of clinical symptoms that are more directly related to NFT and synapse dysfunction and loss. On the basis of animal models and association studies in humans, the immune system is likely to be one of the critical contributors to tangle formation and synapse dysfunction through inflammatory processes and modification of APP processing.
In vitro studies and studies in animal models have provided evidence for the molecular and cellular mechanisms possibly involved in the contribution of the immune system to AD pathology. Until now most efforts to understand the role of the immune system have focused on inflammation and the role of activated microglia.
The immune network depends on specialized cells in every tissue or organ to provide surveillance for potentially foreign agents, to signal to the rest of the immune system cells their presence, and to remove debris. Microglia are brain macrophages that function in phagocytosis, recruitment of T cells, and presentation of antigens (82). At rest these highly ramified cells can surveil up to 50 μm of extracellular brain tissue with some overlap between microglia through a continuous extension and retraction of processes at speeds up to 1.5 μm/min. When activated microglia release complement proteins and inflammatory cytokines, such as IL-1 and IL-6, chemokines, reactive oxygen species, nitric oxide, TNF-α and IL-1β, and matrix metalloproteinases. These released products have been demonstrated to have a direct role in neural damage in cocultures of activated microglia and neurons as well as in AD mouse models (30, 34, 83, 84).
Neurodegeneration as, for example, occurs in the AD brain can cause microglia to be activated. Microglia can also be directly activated by the binding of Aβ (85, 86) as well as by the costimulatory molecules CD40/CD40 ligand (CD40L), found to be increased in AD and animal models of AD (87, 88). Further, CD40/CD40L have been shown to promote pathological changes (amyloid load, gliosis, and hyperphosphorylation of tau) in animal models of AD and, along with IL-1, promote the up-regulation of APP expression (87, 89—92). Aβ, however, is also capable of directly activating the complement system through the classic and alternate pathways by binding C1q and C3b (30, 86). The activated complement system is capable of inducing damage to healthy tissue. Moreover, degenerating neurons and oxidative stress induced by Aβ may also serve as secondary triggers of inflammation with the inflammatory process itself triggering increases in apolipoprotein E and α-1-antichymotrypsin that, in turn, can promote Aβ deposition (93, 94).
Although inflammation in the brain is generally considered harmful because of the damage it can do to healthy cells, data suggest that some components of the inflammatory process may be protective in the amyloid-burdened brain. For example, in an AD rodent model with increased expression of a soluble complement receptor-related protein y that inhibits complement activation, there is a marked increase in Aβ accumulation (95). Further, microglia can secrete enzymes that degrade Aβ (96, 97) and can also phagocytose Aβ (85, 98—102). Additional support for a protective role for microglia derives from experiments in which lipopolysaccharide infusions have increased amyloid clearance in an AD mouse model, presumably as the result of increased phagocytosis (103). Similarly, one likely mechanism whereby active and passive immunization against Aβ (see below) clears brain amyloid is through an increase in microglia phagocytosis (2, 3, 105).
This diversity in action of activated microglia has led to the idea that microglial activation exists on a continuum from "innate activation," characterized by phagocytic response on one end, and "adaptive activation," characterized by antigen-presenting cell function with expression of class II major histocompatability complex (MHC II) proteins on the other. Where along the continuum activation occurs is determined by the specific stimulatory and costimulatory environment (105—110). For example, the above mentioned CD40-CD40L interaction leads to adaptive activation and away from innate activation. Th1 type "regulatory or anti-inflammatory cytokines" (IL-4, IL-10, and transforming growth factor-β) also tip activation of microglia toward the antigen-presenting cell continuum, whereas Th2 type "proinflammatory cytokines" (interferon-γ, TNF-α, and IL-6) shift the activation back toward innate activation. It is important to note, however, that although activation by Aβ1—40 or lipopolysaccharide can lead to increased phagocytosis, the activated microglia can also be toxic to neurons and impair neural cell renewal (111).
Microglia are pivotal cells in the immune response to abnormal APP processing with studies suggesting that activation of microglia may occur along more than one dimension. Further complicating the issue is the fact that the nature of the microglial response may depend on the brain region involved, the amount and rate at which neuronal damage is occurring, the condition of the blood-brain barrier, and the presence of other factors influencing the innate immune system. Moreover, it is possible that the senescence of microglia as individuals age that results in a reduced capacity of microglia to phagocytose amyloid as well as secrete neuroprotective growth factors may contribute to both age-related cognitive changes as well as to AD risk (87, 112, 113). Unfortunately, very little is known about the effect of age on microglial function in AD animal models and even less about the role of astrocytes and T cells, the two other common cell types that are part of the brain's immune system.
Although the number of T cells that enter the healthy brain is substantially less than that in other organs, T lymphocytes that have been stimulated to the blast phase are capable of entering the central nervous system (CNS) presumably in search of antigens. Inflammation or tissue damage enhances neural entry of T cells (114—116). T cells that enter the healthy CNS find an environment that is not conducive to their survival although the mechanisms are unknown. However, if presented with antigen, T cells in the CNS are capable of accumulating. One of the most exciting recent findings has been the discovery of the capacity of activated T cells to secrete neurotrophic factors that foster repair of damaged tissue in the CNS. This well-controlled antiself-response may help the body resist neurodegeneration, i.e., a form of protective autoimmunity (106, 107). The response has been demonstrated to be mediated by autoimmune T cells that produce cytokines and growth factors that activate microglia to produce growth factors that are neuroprotective, e.g., insulin-like growth factor-1. IGF-1 increases Aβ removal through the choroid plexus, stimulates neurogenesis, and through a variety of intracellular pathways inhibits glycogen synthase kinase 3β, an enzyme that plays a role in phosphorylating tau (117, 118).
Three new treatment approaches recently evaluated in AD mouse models emphasize the critical role of T cells in regulating the immune system response as well as microglia activation in the amyloid-burdened brain. Ethell et al. (119) transferred T-cell enriched populations of Aβ-specific immune cells from nontransgenic littermates into cognitively impaired AD mice. The adoptive transfer reversed cognitive decline and synaptic loss for at least 2.5 months after a single infusion. Butovsky et al. (120) exposed T cells in vitro to glatiramer acetate, a U.S. Food and Drug Administration-approved synthetic copolymer resembling aspects of myelin used in the treatment of multiple sclerosis, before infusing them back into an AD mouse model. The reduction in amyloid load and restoration of memory in the treated animals was associated with a switch in activated microglia from a CD11b+/TNFα+ type to a CD11c+/MHC II+ type with dendritic-like morphology, resulting in a lowering of amyloid load and restoration of cognition. It now appears, however, that these are dendritic cells derived from the bone marrow (121). Based on the previous work of these authors, these cells appear to be capable of expressing antigen, secreting growth factors, and buffering glutamate (106, 107, 122). In contrast to the direct infusion of T cells, Frenkel et al. (123) used a proteosome-based adjuvant plus glatiramer acetate vaccination strategy to presumably increase T cell entry in an AD mouse model. Among animals treated, less amyloid load corresponded to greater numbers of T cells, interferon-γ secreting cells, and innate activated microglia.
Perfusions of CNS antigen-exposed T cells or T cells from healthy animals lower amyloid load and restore memory in AD mouse models. Although the above studies emphasize the potential therapeutic importance of T cell infusion or modification, these studies do not directly address the role of "native" brain T cells in the immune response to amyloid pathology or its age dependency. Because microarray studies of aging in humans have consistently shown increased expression in genes associated with inflammation in the brain (124, 125) and other studies have shown increased numbers of activated astrocytes and microglia (126), the aging of the brain's immune response to β-amyloid provides one plausible explanation for the strong age dependence of AD risk.
In addition to the earlier mentioned nonsteroidal anti-inflammatory drugs, epidemiological studies have also suggested that a reduction in AD risk is seen in individuals who are receiving statins. These inhibitors of 3-hydroxy-3-methylgutaryl coenzyme A reductase, thought perhaps to be acting through a mechanism related to cholesterol and its effect on membrane structure and Aβ processing, might also have an immune system component by down-regulating MHC II molecules and costimulatory molecules and cytokines that favor Th1 T-cell development, moving T cells toward a Th2 response (126). Further, in AD mouse models Aβ40 and Aβ42 are able to attract bone marrow cells into the CNS, resulting in microglia that can phagocytose β-amyloid (128). These cells could potentially be modified to express various factors including enzymes capable of degrading amyloid and neurotrophic and neuroprotective factors (108).
Although passive and active vaccination approaches that target amyloid have attracted the most attention, there remain issues related to efficacy as well as concerns about potentially serious side effects that include encephalitis and microhemorrhages (128). Recently Obregon et al. (129) reported the use of a dual active vaccination strategy in mice in which active vaccination against Aβ is combined with active vaccination against CD40L, potentially resulting in a reduction in cerebral amyloid angiopathy and microhemorrhages.
Strategies for the prevention and treatment of AD run the gamut from macromolecular, e.g., cognitive stimulation, to the molecular; most molecular approaches have focused on targets based on the "amyloid cascade hypothesis" (4). These approaches have included
The immune system represents an attractive target for both enhancing clearance of Aβ and providing neuroprotection. However, we are only in the early stages of understanding of the role of the immune system in aging and AD, having created the beginning of an inventory of the responses of some of the elements of the immune system, but still far from a blueprint. To be successful it will be necessary for us to decipher how and when different cells of the brain's immune system work together, how messages travel between and within the cells of this immune network, and how the effects of aging and the elements of AD pathology spread along this complex immune network and are modified by different genotypes. Despite these challenges, it appears likely the immune system will be an important component of any future strategy to treat neurodegenerative disorders.