Glial Cells in Alzheimer’s Disease
4.2. ALZHEIMER’S DISEASE
Brain Amyloids and Other Aggregates of A` Peptides
A diagnostic characteristic of AD is a sufficient density of extracellular senile plaques (SPs) that contain the amyloid ` peptide (A`) in certain brain regions. The A` peptides of up to 43 amino acid residues are endoproteolytically derived from the amyloid precursor protein (APP).
Senile plaque amyloids consist mainly of A` 1–42 but with some longer and shorter peptides, whereas cerebral blood vessels accumulate amyloid containing the slightly shorter A` 1–40. Another diagnostic of AD is the accumulation of intraneuronal aggregates of hyperphosphorylated tau, which are described as neurofibrillary tangles (NFTs). APP, A`, and tau are widely produced by cells throughout the body. From one perspective, the puzzle of AD is to understand why proteins that are present throughout life in body fluids (A`) or within neurons (tau) aggregate during aging. We note briefly that the accumulations of aggregated A` and hyperphosphorylated tau are extremely common during aging of primates and most other mammals that live longer than 10 yr (26,88). These and other species-generalized aging changes define a canonical pattern of aging in mammals (24).
We emphasize that many other forms of aggregated A` 1–43 peptides are found widely during aging in many brain regions. These heterogeneous extracellular materials range widely in morphology and binding of Congo red, which is a required criterion for designation as an ‘‘amyloid.’’ At one extreme may be oligomeric forms of A` (“ADDLs”), which would not be detected by the usual aqueous immunocytochemisty because of their solubility (60). A higher level of A` aggregation is represented by the amorphous, or diffuse, A` deposits detected by immunohistochemistry to A` peptides, but do not distinctly bind Congo red, and hence, are not called amyloids (3,136). The highly compact, Congo red binding A`-containing
deposits of senile plaques and cerebral vessels are the classic amyloid of AD brains. Because of neurons with abnormal dystrophic neurites (swollen, twisted) that are nearby or growing though their matrix, senile plaques are also called neuritic plaques. Another type of deposit in AD brains is described as a “fleecy amyloid” (114). Although the amorphous deposits may arise before the senile plaques, there is no information on causality.
The apparent rapidity with which brain amyloids can form after head trauma (30) indicates the need for detailed studies on the time-course of amyloidogenesis, which may be forthcoming from mouse models engi-neered with human AD genes.
The significance of these diverse A`-containing deposits to AD patho-genesis is highly controversial. On one hand, Terry, Masliah, and colleagues of the UC San Diego Alzheimer Center have emphasized that in the cerebral cortex, the amyloid load is less strongly correlated with the degree of clinical dementia than the synaptic loss (112,113). During the course of clinically demonstrated AD, the total amyloid load does not change (46). On the other hand, Cotman and colleagues of the UCIrvine Alzheimer Center have shown strong correlations of cognitive functions with the total amyloid load as determined in AD brains at various stages (16) and in dogs (17). Moreover, careful examination of the neocortex at very early stages of AD from the Washington University (St. Louis) Alzheimer Center showed all those with
“minimal cognitive dysfunction” had many neuritic plaques, whereas cognitively normal individuals of the same age had a much lower density of amyloid deposits (76). Diffuse plaques were found in all brains and, even in the early AD, they were fivefold more common than neuritic plaques.
Although some authors emphasize that neurons tend to have normal morphology around diffuse plaques with loss of synapses (112), others have observed a smaller cholinergic neuron fiber density in nondemented elderly with diffuse A`-containing deposits, also consistent with early pathogenesis (7).
A large body of work on the role of A` in AD has focused on the neurotoxicity of fibrillar A`. Complex aggregates form rapidly during incubation of various A` peptides (A`1–40, A`1–42, A`25–35) at ambient temperatures; these high-molecular-weight aggregates have widely varying toxicity (102). However, work from this laboratory in collaboration with Klein and Krafft of Northwestern University has demonstrated that oligo-meric (soluble) A` aggregates are highly toxic to neurons (60,81).
Transgenic mice that overexpress human AD genes and have increased production of A` peptides also show neuronal dysfunctions in the absence of A` deposits (41,42,43,89,107), which implies a role for small A` aggre-gates. The neurotoxic pathways involve oxidative stress (8,60,81) and appear
to be mediated by signaling systems with Fyn and Rac 1 (Longo and Finch, unpublished). Superoxide and redox-active iron are both implicated as mediators of A` neurotoxicity (8; Longo and Finch, unpublished).
Inflammatory Processes in Amyloid Aggregates
Besides the neurotoxicity of various forms of aggregated A` peptides, they may participate in inflammatory mechanisms at many levels (Table 1).
By inflammatory mechanisms, we mean to include several subsets of the cellular and molecular changes observed in injured peripheral tissues. We note several major exceptions to inflammatory processes in the brain during AD from those of peripheral inflammation: (1) the absence of swelling; (2) the absence of pain (the brain parenchyma is unique in its paucity of nociception); and (3) the scarcity of B- and T-cells, thereby sharply distinguishing AD from multiple sclerosis in which autoreactive T-cells have the major role in pathogenesis. With its ‘‘cold’’ variety of inflammation, the brain may offer a unique opportunity to study complement functions independently of B- and T-cells. Finch and Marchalonis (25) proposed that AD is a model for the evolutionarily early stages of inflammatory mecha-nisms that preceded combinatorial cellular mechamecha-nisms in immune responses.
A prominent cellular change during AD is the activation of microglia (15,19,22,72,86,91). Microglia are bone-marrow-derived cells of the mono-cyte lineages that, like peripheral tissue macrophages, become phagocytic and produce reactive oxygen species. Glial activation has been recognized from the beginning of AD science: activated glia were described in “pre-senile dementia’’ brains by Alzheimer in 1907 (4), whereas microglia were observed near fibrous A` in senile plaques by Terry et al. in 1964 (111).
See Chapter 3 for more details of glia in senile plaques. In general, fewer activated microglia are associated with diffuse A` deposits (15,87,93).
A continuum of aggregation states is indicated, in which increased micro-glial and astrocyte activation parallels the intensity of thioflavin-S staining (15,87). Of particular interest, the activation of microglia appears to preceed that of astrocytes. During AD, astrocytes also become activated, as gener-ally evaluated by the increase of cellular extensions containing GFAP, the intermediate filament. It is widely recognized that GFAP expression increases in response to local brain injury (64). Moreover, we observed that systemic pathophysiology can stimulate GFAP expression (e.g., in associa-tion with wasting diseases and pathology of non-neural organs) (34).
It is now clear that A` peptides can activate microglia/monocytes and astrocytes (Table 1), the latter including oligomeric A` forms (A`<derived diffusible ligands [ADDLs]) (45). A` also stimulates astrocyte production of interleukin-1 (IL-1) (31,45). Moreover, A` can directly activate the
classical complement cascade by binding to C1q, which is the initial component of the classical complement cascade (Table 1). A general hypothesis is being considered in AD research, by which aggregated A`
initiates inflammatory responses (2,15,22,25,91).
Many inflammatory proteins are detected in senile plaques including cytokines, complement factors, and acute-phase proteins (Table 2). A general note of caution is that these histochemical observations are semiquantitative at best and are sensitive to fixation and to the specificity of the antibodies.
Of great interest to inflammatory mechanisms, C1q shows strong immunostaining in senile plaques (1,22,91). Activation of C1q can produce the anaphylactic peptides (C3a, C4a, C5a), which are chemoattractants and which, like C1q itself, can stimulate oxygen bursts (20). The complement cascade can culminate in production of the cytocidal membrane attack complex (MAC), which contains C5b–C9. Although MAC components and MAC inhibitors are found in AD brains (137,141), there is no information on their role in neuron death during AD.
The amorphous/diffuse deposits of A` appear to have fewer inflamma-tory components. C1q immunohistochemical signals are markedly less in diffuse plaques of the same brains, which show strong signals in neuritic plaques (1,22). However, C3d, apoE, and apoJ are regularly detected in diffuse/amorphous plaques and neuritic plaques of affected brain regions (142). Ongoing work from this lab indicates the presence of inflammatory markers in very early stages of AD (140), the CDR of 0.5 or minimal cogni-tive impairment (76).
In contrast to the robust indications of inflammatory processes in the AD brain, several cytokines in the cerebrospinal fluid (IL-1 , IL-1ra, IL-6, tumor necrosis factor [TNF]) did not change during the dramatic and rapid brain atrophy observed by brain imaging during longitudinal studies of the same patients from the Oxford Aging and Alzheimer Center (63). These findings
Table 1
Pro-inflammatory Activities of A` Peptides Astrocytes
NF-gB, NO, d IL-1
GFAP fibril thickening (stellation) (45) Microglia
Immunoepitopes of activation Respiratory bursts (6,121) Interactions with complement
Binds and activates C1q (122,125)
agree with studies of these cytokines, which were based on single-time cerebrospinal fluid CSF sampling (29,68). However, other CSF inflamma-tory markers may change markedly during AD: on one hand, the level of C1q varied inversely with the clinical rating, which implies its consumption by complement activation (105). However, IL-6 and its soluble receptor in CSF do not change consistently during AD (68). We also note the increased expression of complement genes in other neurological diseases indepen-dently of AD-like amyloid deposits. For example, we observed increased C1q and apoJ in sporadic amyotrophic lateral sclerosis (35). Because C1q may be activated by many components of neurodegeneration, including Table 2
Neuroinflammatory Changes in Alzheimer and Aging Brain Alzheimer’s disease Normal Normal senile plaquesa aging rodentb aging human
Astrocytes X X(GFAP) X (GFAP)c
Microglia X X (OX-6, -42) X
Neurite abnormalities X (NFT) X (no NFT)
A` X X
APP X
_1-antichymotrypsin X _2-macroglobulin X
apoE X X (mRNA)
apoJ (clusterin) X X (mRNA) Corpora amylaceae
CRP X
Heme oxygenase-1 X X (ICC)
Complement factors
C1q X, decrease in CSFf X (mRNA)
C3 X Increase in CSFd
C9 X Corpora amylaceaee
Cytokines
IL-1 X
IL-6 X Plasmag
TGF-`1 X X (mRNA)
aReferences 22, 25, and 137, and Chapter 3.
bReferences 75 and 86.
cReference 79.
dReference 67.
eReference 103.
fReference 105.
gReferences 13, 18, and 92.
myelin and DNA released from dying cells, there may be multiple steps in AD which involve inflammatory mechanisms.
The sources of inflammatory proteins in AD brains are poorly understood.
The skeptic would simply dismiss these findings as a postmortem artifact of blood-brain barrier breakdown during death. However, immunoglobulins are not found in the same senile plaques, which present certain other serum proteins (22,91). Local brain cells are a major potential source of inflamma-tory proteins in association with A` aggregates. Our laborainflamma-tory was the first to show by in situ hybridization that C1q mRNA is relatively abundant in neurons and microglia of human and rodent brains (61,62,83,94). Moreover, we and others (Patrick and Edith McGeer, University of British Columbia;
Joseph Rogers, Sun Health Research Insititute, Sun City AZ) in 1991–1993 reported that C1qB mRNA increases several fold in frontal cortex of AD brains (49,61,62,83,123). Consistent with these findings, we also detected C1q immunoreactivity in surviving hippocampal CA1 pyramidal neurons after excitotoxin lesions (94). At the time, these findings were viewed very skeptically. However, others soon showed C1q immunostaining in neurons of AD brains (1) and increased C1q mRNA in AD brains (28). Recently, we showed that rat brain can synthesize de novo bioactive C1q during responses to lesions (32). Evidence from many sources indicates that brain contains mRNA and proteins that represent most, if not all, classical and alternate path complement components, including the C9 of the membrane attack complex (e.g., ref. 137). The multiple functions of C1q include intracellular activities (binding to calreticulin) as well as interactions with a wide range of other systems that mediate normal tissue renewal (20). Moreover, we observed that C1q mRNA is expressed during brain development in association with regional synaptogenesis (50). We also briefly note that some of the inflamma-tory proteins associated with aggregated A` can also modify the activities of glial cells (e.g., apoE attenuates activation of astrocytes by A`) (44), whereas apoJ (clusterin) activates microglia (135).
4.3. INFLAMMATORY MECHANISMS IN BRAIN AGING