Interleukin-1 and the Immunogenetics of Alzheimer Disease

Abstract

Established genetic causes of familial Alzheimer disease (AD) involve genes for β-amyloid precursor protein (βAPP), presenilin-1, and presenilin-2. For the more common sporadic forms of AD, increased risk has been associated with a number of genes; the most important of which is the ε4 allele of apolipoprotein E. Two recent studies, one clinical and one using postmortem material, now show increased risk for AD associated with certain polymorphisms in the genes encoding the α and β isoforms of interleukin-l (IL-1). IL-1 levels are elevated in Alzheimer brain, and overexpression of IL-1 is associated with β-amyloid plaque progression. IL-1 interacts with the gene products of several other known or suspected genetic risk factors for AD, including βAPP, apolipoprotein E, α₁-antichymotrypsin, and α₂-macroglobulin. IL-1 overexpression is also associated with environmental risk factors for AD, including normal aging and head trauma. These observations suggest an important pathogenic role for IL-1, and for IL-1-driven cascades, in the pathogenesis of AD.

INTRODUCTION

Alzheimer disease (AD) occurs in both familial and sporadic forms. Rare pedigrees showing early onset and dominant inheritance have allowed identification of pathogenic, or causative, mutations in genes for the β-amyloid precursor protein (βAPP) on chromosome 21 (1), for presenilin-1 on chromosome 14 (2), and for presenilin-2 on chromosome 1 (3).

For the more common sporadic (and usually late-onset) forms of the disease, increased risk has been reported for a number of genes, the most important of which is the ε4 allele of apolipoprotein E (ApoE) (4). However, possession of this ApoE allele is neither necessary nor sufficient for the development of AD, and other genetic factors have been sought. Additional candidate genes include those encoding α₂-macroglobulin, an acute-phase reactant protein that is co-deposited in β-amyloid plaques; the very low-density lipoprotein receptor, which binds both apolipoprotein E and α₂-macroglobulin; butyrylcholinesterase, an enzyme of unknown physiological function that is overexpressed in brain of Alzheimer patients; and α₁-antichymotrypsin, another acute-phase reactant, also co-deposited in the β-amyloid plaques of AD.

For the α₂-macroglobulin gene, a splice acceptor deletion has been associated with increased risk of AD (5). For the very low-density lipoprotein receptor, a polymorphism containing 5 trinucleotide tandem repeats has been reported to confer risk among Japanese (6), but this increased risk is not seen among Caucasians (7). An increased risk associated with an allelic variant of the butyrylcholinesterase gene, involving an alanine to threonine amino acid substitution, has been suggested (8) but not confirmed (9, 10). For the α₁-antichymotrypsin gene, 2 distinct associations with AD have been suggested. A common polymorphism in the signal sequence may interact with ApoE ε4 alleles to either increase risk (11) or to lower disease onset age (12), but this association was not confirmed (13). In another study, a polymorphic dinucleotide repeat in the 5′ flanking region of the α₁-antichymotrypsin gene was reported to act in association with ApoE ε4 as a risk factor for AD (14). This dinucleotide polymorphism showed strong linkage disequilibrium with the signal sequence polymorphism mentioned above, and thus the association of the flanking region polymorphism with AD contradicts the reported association of the signal sequence polymorphism with AD. In addition to these candidate genes, a biallelic polymorphism (but not a mutation) in the presenilin-1 gene has also been associated with sporadic, late-onset AD (15).

These identified genes do not fully explain genetic risk for AD, suggesting that additional genetic influences remain to be elucidated. In addition, these noncausative but risk-conferring genes must interact with environmental risk factors, which include increasing age (16) and a history of significant head trauma (17), as well as the negative risk factor (or protective effect) of anti-inflammatory drug use (18). The latter observation, in particular, correlates closely with extensive work showing a significant immunological (or “inflammatory”) component in Alzheimer pathogenesis and in disease progression (19, 20), and suggests that genes encoding proteins involved in inflammatory pathways might also modulate risk for AD. Indeed, the inclusion of 2 acute-phase reactant proteins (α₂-macroglobulin and α₁-antichymotrypsin) among previously identified candidate genes for AD supports this suggestion.

GENETIC RISK ASSOCIATED WITH INTERLEUKIN-1 POLYMORPHISMS

Two recent studies, one based on clinically assessed patients and the other on neuropathologically confirmed cases, have shown increased risk for AD associated with certain polymorphisms in the genes encoding the α and β isoforms of interleukin-1 (IL-1), independent of ApoE genotype (21, 22).

IL-1 is a potent pro-inflammatory (“acute phase”) cytokine, with numerous systemic effects (23), that is overexpressed in brains of Alzheimer patients. There are 2 isoforms of IL-1, α and β, that are encoded by separate genes (IL-1A and IL-1B, respectively), located on the long arm of chromosome 2 in a cluster with the gene for the IL-1 receptor antagonist (24). Both isoforms are synthesized as 33 kDa precursors that are cleaved to yield 17 kDa products. For the 13 isoform, only the secreted 17 kDa cleavage product is biologically active, whereas for the α isoform both the nonsecreted 33 kDa and the secreted 17 kDa molecules are biologically active. Microglia are the principal source of both IL-1α and IL-1β in the central nervous system (25-27), although some production also occurs in astrocytes (28) and even in neurons (29, 30).

The IL-1A gene contains a common polymorphism in the 5′ regulatory region (a C to T transition at position –889 relative to the start site of transcription) that results in 2 alleles, allele 1 and allele 2. IL-1A allele 2 has been associated with juvenile rheumatoid arthritis, suggesting a functional effect of this polymorphism on inflammatory processes (31). The IL-IB gene contains a common polymorphism in the coding region that introduces a Taq1 restriction site. This results in 2 alleles, designated allele 1 and allele 2 (32, 33). Homozygosity for allele 2 induces a 4-fold elevation in IL-1β production, compared with IL1-B 1/1 homozygotes (32). In addition, individuals with a composite genotype consisting of both IL-1A allele 2 and IL-1B allele 2 carry increased risk for periodontitis (33), providing further evidence that these genetic polymorphisms modulate inflammatory processes.

In a study of 232 neuropathologically confirmed cases of AD and 167 controls from 4 participating centers on 2 continents, Nicoll et al (21) found the IL-1A 2/2 genotype present in 13% of Alzheimer patients but in only 7% of controls, yielding a tripling of risk for AD. Even greater risk was associated with a composite IL-1A 2/2, IL-1B 2/2 genotype: Only 17 of 399 patients carried this genotype, but 15 of these 17 had AD, yielding a 10-fold increase in risk. In an Italian study of 318 patients with clinical diagnoses of AD and 335 controls, Grimaldi et al (22) found a doubling of risk for AD associated with the IL-1A 2/2 genotype. This increased risk was attributable to a 4-fold increased risk for earlier disease onset, as those Alzheimer patients carrying the IL-1A 2/2 genotype became symptomatic 7 to 9 years earlier (61 years) than those with either IL-1A 1/2 (68 years) or IL-1A 1/1 (70 years) genotypes. This study also found a small increase in risk for AD, albeit for a late age at onset, associated with another polymorphism present in the promoter regions of IL-1B.

These 2 studies, both showing increased risk of AD with IL-1 gene polymorphisms that have been previously associated with increased IL-1 production and with other inflammatory diseases, underscore the role of IL-1 and of immunological (“inflammatory”) processes in the pathogenesis of AD.

INTERLEUKIN-1 IN ALZHEIMER DISEASE

In contrast to the uncertainty surrounding the pathophysiological roles of other identified Alzheimer risk-conferring gene products, there is considerable evidence for an important pathophysiological role for IL-1 in AD. For instance, IL-1 regulates βAPP production and processing, and direct injection of IL-1 into brain results in neuronal overexpression of β-APP (34). Chronic elevation of intracerebral IL-1 levels, achieved by implantation of IL-1-impregnated, slow-release pellets into rat brain, results in increased expression of, and increased phosphorylation of, both neurofilament and tau proteins (35). These direct demonstrations of neural injury associated with increased intracerebral levels of IL-1 contrast with the paucity of evidence for direct in vivo neurotoxic effects for β-amyloid (36, 37).

In AD, IL-1 levels are elevated in both brain tissue (38) and cerebrospinal fluid (39). Activated, IL-1-immunoreactive microglia in the cerebral cortex in AD are 6-fold more numerous than those in control brain (38). These microglia show patterns of distribution across brain regions that correlate with the distribution of β-amyloid plaques (40), and they are frequently found within these plaques. Even the early, diffuse (nonfibrillar and nonneuritic) “pre-amyloid” deposits contain activated microglia, overexpressing IL-1, in patients with AD (41), in contrast to a lack of microglia in the similar diffuse amyloid deposits sometimes found in nondemented elderly individuals (42). These observations suggest that there is some chemical or physical difference between the diffuse “pre-amyloid” deposits in AD and those seen in normal aging that renders the diffuse deposits of AD both immunogenic and pathogenic. In AD, plaque progression, which is evident as condensation of amyloid to form β-amyloid and as formation of dystrophic neurites, is accompanied by increases in the number, size, and IL-1 immunoreactivity of plaque-associated microglia (41). As a consequence, the number of plaque-associated microglia waxes and wanes through an hypothesized sequence of plaque evolution, from sparse microglia in diffuse non-neuritic plaques, to abundant microglia in diffuse neuritic plaques, to mildly decreased numbers of microglia in dense-core neuritic plaques. End stage (“burned out”) plaques, consisting solely of dense β-amyloid cores without associated diffuse amyloid or associated neuritic elements, are devoid of microglia, suggesting loss of immunogenic potential in these end-stage lesions. Neurons within or adjacent to β-amyloid plaques show evidence of injury in the form of DNA damage demonstrated using the TUNEL technique (43). The extent of injury to plaque-associated neurons, as indicated by the relative numbers of TUNEL-positive neurons, increases progressively with plaque stage and, presumably, with continued exposure to locally elevated levels of IL-1. End stage plaques show markedly diminished numbers of plaque-associated neurons, and all of those neurons label with TUNEL, suggesting that there is also neuronal loss during plaque evolution as a consequence of IL-1-mediated neuronal injury (43). These observations collectively suggest an important pathogenic role for activated microglia and for IL-1 in neuritic plaque formation and progression and in the accompanying neuronal injury and loss in AD.

INTERLEUKIN-1 AND OTHER GENETIC RISK FACTORS FOR ALZHEIMER DISEASE

Interleukin-1 is known to interact with the gene products of several other known or suspected genetic risk factors for AD. For instance, IL-1 promotes both neuronal synthesis (44, 45) and processing (46) of βAPP, with potential contributions to neurotoxicity associated with either secreted βAPP fragments (47) or possibly with β-amyloid itself. Conversely, secreted βAPP α fragments promote microglial activation and overexpression of IL-1 (47). This latter effect is modulated in an isoform-specific manner by ApoE: that is, ApoE ε3 blocks this effect while ApoE ε4 does not. These observations suggests that the increased risk of AD associated with ApoE ε4 might derive in part from inability of this ApoE isoform to inhibit the pro-inflammatory effects of secreted βAPP α fragments. The possible clinical relevance of these interactions is supported by the observation that sera from ApoE 3/4 and ApoE 4/4 Alzheimer patients, but not from ApoE 2/3 or ApoE 3/3 Alzheimer patients or from controls, promote IL-1 release from microglia (48). Further IL-1-A poE interactions are suggested by the observation that IL-1, in combination with other pro-inflammatory cytokines, inhibits ApoE secretion from high-passage cultured astrocytes (49).

Interactions between IL-1, α₁-antichymotrypsin, and α₂-macroglobulin have also been reported. IL-1 has been shown to induce astrocytic expression of α₁-antichymotrypsin (50) and, conversely, α₁-antichymotrypsin has been shown to upregulate astrocytic production of IL-1 (51). α₂-Macroglobulin is a major plasma IL-1 binding protein (52, 53), and IL-1 has been shown to stimulate α₂-macroglobulin synthesis (54).

Down syndrome must also be considered a genetic risk factor for AD, as the association of Down syndrome with florid, early-onset Alzheimer-type neuropathological changes is well known (55). We have shown glial inflammatory changes, including a profusion of activated microglia overexpressing IL-1 in young and even fetal Down's tissue (38, 56, 57). These changes precede by decades the appearance of classic Alzheimer-type neuropathological changes.

INTERLEUKIN-1 AND ENVIRONMENTAL RISK FACTORS FOR ALZHEIMER DISEASE

Microglial activation and overexpression of IL-1 is found in a number of diseases and conditions that predispose to AD or to premature (accelerated) appearance of Alzheimer-type “senile” neuropathological changes. These include increasing age and head injury (which carry elevated risk for AD), and chronic, intractable epilepsy, which is associated with accelerated appearance of Alzheimer-type “senile” changes.

Increasing age is the major nongenetic risk factor for AD. Normal aging is characterized by increases in the size of and IL-1 overexpression by microglia, and by increases in tissue levels of IL-1 mRNA in brain (58). Indeed, there is evidence that the normal distribution of microglia in cerebral cortical layers may determine in part the distribution of neuritic plaques in AD (59). This normal age-related effect may explain the experimental observation that intracerebral injections of purified β-amyloid are neurotoxic in aged, but not young, primates (37).

Head injury is an important environmental risk factor for AD (17). Acute, severe head injury is accompanied by increased neuronal expression of βAPP and increased microglial expression of IL-1 (60) and, in approximately one third of patients, by cerebral cortical deposition of β-amyloid protein (61). Indeed, plaque-like clustering of dystrophic neurites, invariably associated with activated microglia overexpressing IL-1, can be found even in young head injury victims (60).

Epilepsy is not an established risk factor for AD, but patients with chronic intractable epilepsy do show an accelerated appearance of Alzheimer-type “senile” neuropathological changes (62) that is most striking in patients that carry the Alzheimer-associated ApoE ε4 allele (63). We have shown increased numbers of activated microglia overexpressing IL-1 and elevated neuronal expression of βAPP in temporal lobe specimens from patients with chronic intractable epilepsy, further suggesting that IL-1 is important in the pathogenesis of Alzheimer-like pathological changes (64).

INTERLEUKIN-1 AS A KEY ORCHESTRATING CYTOKINE IN ALZHEIMER DISEASE

The increased genetic risk for AD that obtains with specific IL-lα and IL-1β genotypes is consistent with an important pathogenic role for IL-1 and IL-1-driven cascades in the appearance and progression of lesions in Alzheimer brain. IL-1 has been shown to have a number of important effects on neurons, on astrocytes, and on microglia that are relevant to Alzheimer pathogenesis. These effects include overexpression of additional trophic molecules of pathogenic import with further cascades of molecular and cellular events. We have proposed that conditions and insults that engender chronic overexpression of IL-1, sustained at sufficient levels and over sufficient periods of time, can initiate a self-propagating cascade of detrimental cellular and molecular events known as the cytokine cycle. This cycle culminates in the progressive neuronal injury and loss as well as the progressive neurological decline associated with AD (20, 65).

Specific neurotrophic effects of IL-1 relevant to AD include promotion of synthesis (44, 45) and processing (46) of βAPP, thus favoring the release of both amyloid peptide fragments (and further deposition of β-amyloid) and gliotrophic (47) secreted βAPP fragments. IL-1 also has effects on neuronal cholinergic functions and neuronal stress reactions. IL-1 directly enhances neuronal acetylcholinesterase expression and activity, and IL-1-mediated glial-neuronal interactions are responsible for increased neuronal acetylcholinesterase activity following glutamate-induced neuronal stress (66). Thus, the well-recognized cholinergic dysfunction of AD may derive in part from the overexpression of IL-1 in Alzheimer brain.

IL-1 acts on astrocytes to promote astrocyte activation and to promote increased astrocytic expression of both S100β (34) and α₁-antichymotrypsin (50), and acts on microglia (autocrine effects) to promote further activation and further stimulation of IL-1 expression (66, 68, 30). Astrocytic overexpression of S100β is a particularly important consequence of increased IL-1 levels. Activated astrocytes are prominent, and tissue levels of biologically active (homodimeric 10 kDa) S100β are markedly elevated, in Alzheimer brain (69). Activated astrocytes, overexpressing S100β, show patterns of association with different types of β-amyloid plaques that parallel the associations of activated microglia, overexpressing IL-1, with these plaques (70). S100β is a neurotrophic cytokine that promotes neurite outgrowth (71) and the extent of astrocyte activation and S100β overexpression correlates with the extent of neuritic swelling and injury in individual β-amyloid plaques (70). S100β also promotes increases in intracellular free calcium concentrations in neurons (72) and induces neuronal βAPP and interleukin-6 production (73). S100β has autocrine effects on astrocytes that include induction of astrocytic nitric oxide synthase activity with release of potentially neurotoxic nitric oxide (74).

The IL-1-initiated cascades inherent in the cytokine cycle include several potentially neurotoxic steps, including raised intraneuronal free calcium concentrations, overstimulation of neuritic outgrowth, and increased tissue levels of nitric oxide. Feedback mechanisms, with further activation of microglia and promotion of IL-1 overexpression, both sustain the immunological process and promote continuing neuronal injury. Neurotoxicity associated with activation of the cytokine cycle may thus drive the neuronal injury and loss associated with progression of AD.