Lessons from genome-wide association studies findings in Alzheimer's disease
Funding: This work was supported by FAPDF (grant # 193.000.449–2008) and Finatec/UnB (grant # 5563/2009) to O.T. Nóbrega. Dr R.W. Pereira was supported by a fellowship for productivity in research from CNPq.
Conflict of interest: none.
Dr Otávio de Tolêdo Nóbrega PhD, Programa de Pós-Graduação em Ciências Médicas, Universidade de Brasília – UnB, Campus Universitário Darcy Ribeiro, Asa Norte, Brasília – DF, 70910-900, Brazil. Email: email@example.com
Alzheimer's disease (AD) is the most common neurodegenerative disorder with a complex genetic background. Recent genome-wide association studies (GWAS) have placed important new contributors into the genetic framework of early- and late-onset forms of this dementia. Besides confirming the major role of classic allelic variants (e.g. apolipoprotein E) in the development of AD, GWAS have thus far implicated over 20 single nucleotide polymorphisms in AD. In this review, we summarize the findings of 16 AD-based GWAS performed to date whose public registries are available at the National Human Genome Research Institute, with an emphasis on understanding whether the polymorphic markers under consideration support functional implications to the pathophysiological role of the major genetic risk factors unraveled by GWAS.
Remarkable increases in life expectancy worldwide have led to a clear rise in the number of cases of neurodegenerative diseases and dementia.1 Dementia syndromes are characterized by a progressive impairment of cognitive function, particularly a decline in memory and interference with social and occupational activities.2
Population-wide epidemiological studies have revealed an increasing incidence of late-onset dementia.3,4 Worldwide, the number of people with some form of dementia was estimated at 24 million in 2000, with this figure expected to double to 42 million within 20 years and then increase to 81 million in 40 years.5 Life expectancy is shortened in people with dementia, and mortality is much higher than in people without dementia (relative risk > 3).6
GENOME-WIDE ASSOCIATION STUDIES
Since 2007, a variety of genetic epidemiology studies have focused on examining the whole genome and assessed hundreds of thousands of single nucleotide polymorphisms (SNPs) to detect genetic variations associated with specific phenotypes. These investigations are now known as genome-wide association studies (GWAS). Recent studies have identified an increasing number of genetic loci known to be involved in the pathogenesis of Alzheimer's disease (AD) and other dementias, and the GWAS approach has gained ever greater prominence in the past few years as a promising tool for investigation of complex phenotypes. The objective of these studies is to test hundreds of thousands of individuals for common polymorphisms with novel associations and to study genes or genomic regions never before regarded as candidates for the etiology of a given condition. Beginning in 2008, the public registry maintained by the National Human Genome Research Institute (http://www.genome.gov/gwastudies; Bethesda, USA) has added 16 GWAS assessing AD to its files,7–22 and thus far, 26 SNPs have been strongly associated (p-value < 1.0 × 10−5) with AD (Table 1).
Table 1. Characteristics of important GWAS SNPs associated with Alzheimer's disease
|CR1*||rs6656401; rs3818361||A/G; C/T||1q32||180941120; 181015472||16|
|CLU*||rs2279590; rs11136000; rs9331888||A/G; C/T; C/G||8p21-p12||26416343; 26424609; 26428952||16|
From a theoretical standpoint, two competing, non-exclusive trains of thought currently attempt to explain the occurrence of complex diseases, also known as ‘common diseases’ when associated with a considerable population-wide frequency.23 The ‘common disease-common variant’ hypothesis maintains that common diseases are caused by genotype or haplotype variants that are common (i.e. those with a population-wide frequency above 5%) but display modest genetic effects and make minor contributions to the disease process. GWAS have proved promising in describing polymorphisms potentially associated with several diseases on the basis of the common disease-common variant hypothesis,24 and they have confirmed classic theoretical associations, such as the finding that the apolipoprotein E (APOE) ε4 allele is associated with increased risk of AD.25
A second theory involves the ‘common disease-rare variant’ hypothesis, which poses that the prevalence and clinical severity of complex conditions can be partly ascribed to the occurrence of multiple rare allelic variants, due to either new mutational episodes or rather frequent events of incomplete penetrance. Several family studies involving, for instance the APP gene, have shown the outstanding effect of rare variants (i.e. those with a population-wide frequency <1%) in determining early-onset forms of AD in inheritance pattern that occasionally may resemble Mendelian inheritance.25,26 The function of these rare variants is occasionally known, and some can produce functional protein mutations. However, because of their low frequencies, identification of these variants can be challenging and is usually accomplished by resequencing previously identified candidate regions.27 GWAS strategies may be useful in these settings.
The present study sought to review the genetic determinants currently known to play a significant role in the pathophysiology of AD, with particular emphasis on the contributions of modern genome-wide studies to the field.
Of the various dementia-causing neurodegenerative conditions, AD is the most prevalent.28 In Brazil, for instance, a study conducted by Herrera et al. found the overall prevalence of dementia to be 7.1% in people over 65 years of age, with Alzheimer's accounting for 55.1% of all cases followed by combined (AD/vascular) (14.4%), vascular (9.3%) frontotemporal (2.6%) and Lewy (1.7%) forms of dementia.29
The most striking changes found in the brains of patients with AD are senile plaques (extracellular deposits of the beta-amyloid, or amyloid beta, protein), neurofibrillary tangles (clumps of hyperphosphorylated tau protein located in the perinuclear cytoplasm) and extensive neuron loss.30
Genome-wide studies have shown that mutations in the amyloid precursor protein (APP), presenilin 1 (PSEN1), and presenilin 2 (PSEN2) genes account for approximately 80% of cases of early-onset AD;30 30% to 40% of these cases are inherited in an autosomal dominant pattern.31 The association between early-onset AD and rare allelic variants has also been described with the APOE and tau protein (MAPT) genes;30,32 extensive evidence of an association between APOE and late-onset AD also exists. However, APP, PSEN1, and PSEN2 mutations only account for roughly 5% of all cases of AD. The remaining 95% constitute, for the most part, sporadic late-onset disease with a complex etiology, due to interactions between environmental and genetic factors.13,31 Recent studies have ascribed a role in the etiology of late-onset AD to several genes other than APOE: clusterin (CLU), phosphatidylinositol binding clathrin assembly protein (PICALM), and complement receptor 1 (CR1).13,16
APOLIPOPROTEIN E (APOE)
APOE, a 317-amino acid glycoprotein with a molecular weight of 34 kDa, is one of the main proteins present in human plasma, to which it is transported after being synthesized in the liver,33,34 and is also the main apolipoprotein found in the human brain. In plasma, APOE is the main component of very low-density lipoprotein and part of the high-density lipoprotein group, which is involved in the redistribution of triglycerides and cholesterol to various tissues.35,36 APOE is one of many distinct classes of apolipoproteins, including APOA, APOB, APOC, APOD and APOJ (clusterin), all of which play a role transporting lipids through plasma and other body fluids.
The nucleotide sequence of the coding region of the APOE gene, located at 19q13.2, is polymorphic; the main alleles are known as ε2, ε3, and ε4. The gene expression products of these alleles are known as APOE2, APOE3, and APOE4 respectively. These products differ by the presence of a cysteine (cys) or arginine (arg) residue in positions 112 e 158: APOE2, cys-112/cys-158; apoE3, cys-112/arg-158; and APOE4, arg-112/arg-158.37 In other words, two SNPs are the determinants of this switch: rs429358 (C/T) and rs7412 (C/T). The haplotypes of these SNPs determine the APOE allele, namely ε2 (TT), ε3 (TC), or ε4 (CC).38 The several possible combinations of two of the three main alleles give rise to six possible genotypes: APOE ε2/ε2, APOE ε3/ε3, APOE ε4/ε4, APOE ε2/ε3, APOE ε3/ε4, and APOE ε2/ε4.
The frequency of these three alleles in populations of different biogeographic origins is highly variable, although it does follow a pattern in which the ε3 allele is always most frequent, occurring in approximately 80% of any population.39 The ε4 allele usually occurs more often than ε2, but in some populations (such as Russians), both occur with equal frequency,40 and in others (such as Han Chinese), ε2 is more common than ε4.41 The ε2 allele has been reported absent from some populations, particularly among indigenous peoples in the Amazon region,42 Mexico,43 and Australia.44 The ε4 allele is relatively more frequent in some African populations and among the indigenous peoples of Oceania and the Sami of Northern Europe.39 A more in-depth review of the population-based distribution of APOE alleles is provided by Corbo and Scacchi.39 Although the frequency of ε alleles varies across distinct populations, the prevalence of AD depends on other genetic and environmental factors as well.
APOE is synthesized by a wide variety of tissues, chiefly in the liver, the output of which accounts for roughly three-quarters of circulating plasma APOE levels.33,34 In humans, the brain is the second most prominent site of APOE synthesis, which mainly takes place in the astrocytes45,46 and microglia.47 As evidence of the specific role of APOE in amyloid plaque and neurofibrillary tangle formation, studies have shown that APOE is one of the components of beta-amyloid plaques in the brain.48 Identification of the APOE ε4 variant as the single most common risk factor for late-onset AD suggests that cholesterol may play a direct role in the pathogenesis of the disease.49 However, structural and biophysical differences in the various isoforms of the APOE protein have distinct effects on the etiology of AD50. Evidence suggests that APOE4 promotes in vivo and in vitro amyloid-protein fibrillogenesis, whereas APOE3 binds the tau protein with greater affinity, reducing its initial phosphorylation rate and blunting filament formation.51,52
The various isoforms of APOE exhibit relative variations in their capacity to induce amyloid plaque formation. The presence of a cys residue in APOE3 and APOE2 leads to the formation of disulfide-bound homo- and heterodimers, which cannot occur with APOE4. This gives APOE4 less molecular stability and makes it more prone to clumping.53 APOE4 is therefore more effective than APOE3 in inducing beta-amyloid formation and deposition.54 In normal individuals, the soluble fraction of APP (soluble N-terminal APP, sAPPα) is completely fragmented by proteases in the extracellular space. In AD, however, experiments suggest that binding to insoluble forms of APOE also leads to deposition and accumulation of amyloid plaques.
Furthermore, individuals with the APOE4 allele are at risk of AD with a disease course affected by SNPs elsewhere in the genome. For instance, the T allele of polymorphism rs1799724 (C/T), on the tumor necrosis factor-alpha (TNF-α) gene appears to synergistically increase risk in the presence of the APOE4 allele and is associated with changes in cerebrospinal fluid levels of Aβ42.55 A specific haplotype of three SNPs on the cell division cycle 2 gene almost doubles the risk of AD,56 whereas allele G of rs2373115 (G/T) on the GRB-associated binding protein 2 gene appears to increase tau phosphorylation.20 Additional evidence is available,57 but should be viewed cautiously, as findings may differ according to the stage of AD or to the ethnicity of the study population, as is the case with GRB-associated binding protein 2 polymorphisms.58–60
AMYLOID PRECURSOR PROTEIN (APP)
APP is a protein expressed in various tissues, but its most significant actions are on regulation and plasticity of neural synapses.61 Regular proteolytic processing of APP, following the non-amyloidogenic pathway, is carried out by the α-secretase enzyme complex (TNF-α converting enzyme, ADAM-9 and ADAM-10), which leads to two products, soluble N-terminal APP and a membrane-bound fragment known as C83.62 The latter product is then cleaved by the γ-secretase complex, composed of PSEN1 or PSEN2,63 nicastrin,64 and APH-1/PEN-2,65 to produce a P3 residue, which is processed in the extracellular space, and an intracellular APP domain known as AICD.62,66
However, in the amyloidogenic pathway, α-secretase does not process APP properly; instead, the β-secretase complex, also known as β-site APP cleaving enzyme I, cleaves APP into soluble APP-β and a membrane-bound fragment known as C99. The γ-secretase complex then cleaves C99 into two fragments, AICD (intracellular) and extracellular, 40- or 42-amino acid-long amyloid beta fragments (Aβ40 and Aβ42), respectively.62 Aβ42 is the neurotoxic fragment of APP and is the major constituent of amyloid fibers and their later formation as senile plaques.
Regulation of APP proteolysis is thus dependent on the activity of a multimeric protein complex that has presenilins, nicastrin, PEN-2, and APH-1 as its main constituents.67 Studies suggest that reductions in APP fragment levels or activity combined with Aβ buildup could play a critical role in the cognitive dysfunction associated with AD, particularly in the early stages of the disease. These data clearly show that APP fragments, including the Aβ42:Aβ40 ratio, may have a powerful regulatory effect on basic neuronal functions, such as cell excitability and synaptic transmission. Therefore, these fragments likely play a role in behavioural regulation, learning, and memory.61
Some genetic variations in the APP gene are the cause of excess Aβ42 production in familial AD. Mutations in the Val717 codon are known specifically to increase Aβ42 secretion, altering the Aβ42:Aβ40 ratio. However, these mutations may be due to two adjacent SNPs, which will determine one of four different types of protein. The first (rs63750264) has three possible variations: a G > A transition, a G > T transversion and another G > C transversion, which lead to the missense mutations Val717Ile,68 Val717Phe69 and Val717Leu,70,71 respectively. Mutation of the adjacent nucleotide (rs63749964) is a T > G transversion that produces the missense mutation Val717Gly.72 Other functional SNPs, such as rs63750066 (Ala713Thr),73,74 rs63750973 (Thr714Ile),75,76 rs63750643 (Thr714Ala),77 rs63750734 (Val715Met),78 and rs63750399 (Ile716Val),79 have also been pinpointed near this codon. These mutations are located in a transmembrane domain-coding region that determines the formation of APP dimers. Modeling of APP has shown that these mutations probably destabilize the APP-transmembrane domain dimer, increasing the ratio of APP monomers. Differential recognition of monomeric and dimeric forms of APP by the γ-secretase complex may thus be related to preferential cleavage at the Aβ40 or Aβ42 sites, with certain point mutations promoting formation of neurotoxic Aβ42.80
PRESENILIN 1 AND 2 (PSEN1 AND PSEN2)
The presenilins genes, PSEN1 and PSEN2, encode multirole proteins that, among other actions, function as the catalytic components of γ-secretase, the aspartyl protease complex that cleaves APP to form amyloid beta.63,81 Several PSEN1 and PSEN2 mutations have been described in early onset familial AD cases.25 PSEN1 mutations account for 18% to 50% of cases of early-onset AD.82 The mode of action of PSEN mutations is debatable, and since their mechanisms are not yet fully revealed, major evidence suggests a loss of function of γ-secretase,67,81,83,84 even though some evidence remains on a gain of function hypothesis.85 For instance, the PSEN1-L166P variant can lead to a global decline in Aβ production, whereas the PSEN1-G384A mutation significantly increases Aβ42 production.83 It is likely that PSEN variants interact differently with proteins in the γ-secretase complex either reducing its activity or changing the cleavage site of APP.83,84,86 Besides, presenilin also plays an important role in other intracellular components' cleavage, such as Notch-1. PSEN mutations were also associated with decreased Notch signaling, thus resulting in an evidence for γ-secretase loss of function.83
PSEN1 and PSEN2 follow distinct patterns of expression in human tissue. Considering that PSEN1 is transcribed uniformly throughout the brain and peripheral tissues, PSEN2 is expressed at relatively low levels in the brain (with the exception of the corpus callosum, where there is substantial expression), but is highly expressed in some peripheral tissues, such as the pancreas, heart, and skeletal muscle.82 Low levels of PSEN2 in the brain and compensatory PSEN1 activity may explain why PSEN2 mutations with incomplete penetrance are a comparatively rarer event than PSEN1 mutations, which are completely penetrant.86,87
PSEN1, located at 14q24.3, plays a role in the inflammatory process observed in amyloid plaques and may interfere with neuronal apoptosis.88,89 PSEN2, located at 1q31-q42, perhaps speeds the neurodegenerative process, depending on the allelic variant.90 In humans, most PSEN1 and PSEN2 gene variants are point mutations, although a small number of deletions and insertions have been reported. Most of these mutations are in the PSEN1 gene (178 mutations in 393 families), with a small portion in PSEN2 (14 mutations in 23 families), which sheds light on the importance of these genes in the pathophysiology of early-onset AD.25 GWAS of late-onset AD have failed to show any association with PSEN1, PSEN2 or APP, which corroborates the role of highly penetrant, rare variants exclusively in early-onset familial AD.
Tau protein is part of the microtubule-associated proteins (MAP) family. The main role of microtubule-associated proteins is to stabilize microtubules by aggregating tubulin.91 In all cell types, microtubules are a dynamic structure that is absolutely essential to the cell division process. In postmitotic neurons, this function is suppressed, and the role of microtubules shifts into maintenance of the cell's architecture and to intraneural transport. Therefore, in addition to their involvement in maintenance of the cell structure and in processes that involve neuroplasticity, microtubules also play an essential role in axonal transport of organelles (mitochondria, endoplasmic reticulum, lysosomes) and vesicles, which, in turn, carry neurotransmitters and proteins from the soma to the distal synapses.92 Tau protein also ensures the uniform orientation of axonal microtubules.93
The human tau protein gene is located on chromosome 17q21 and contains 16 exons. In the human brain, because of alternative splicing, tau protein is present as six isoforms,94 which range from 352 (37 kDa) to 441 (46 kDa) amino acids in length. In healthy nerve cells, all tau isoforms behave as soluble proteins in the axon, whereas in tauopathies, they are widely distributed throughout the soma and dendrites and may be found in soluble or insoluble form. The latter is found in paired helical filaments, the main component of neurofibrillary tangles. In paired helical filaments, six to eight phosphate groups are present for each tau protein unit – a number far superior to the usual degree of tau phosphorylation in the healthy brain (approximately two phosphate groups per molecule). In short, paired helical filaments are composed of tau protein in a hyperphosphorylated state.95
From a functional standpoint, tau controls microtubule dynamics during neurite maturation and outgrowth. As the largest protein in the cytoskeleton, hyperphosphorylated tau has marked effect on the biological functions and morphological features of neurons and jeopardizes tubulin binding capacity, thus destabilizing the microtubules.96 Furthermore, hyperphosphorylated tau is conducive to aggregate formation and blocks intracellular transport of neurotrophic factors and other functional proteins, leading to a loss or decline in axonal and dendritic transport. Thus, hyperphosphorylated tau jeopardizes synaptic metabolism, causing dysfunctions that lead to loss of cell viability, the collapse of the microtubular cytoskeleton, and neuron death.97
Tau protein also promotes the interaction between actin and neurofilaments, which suggests interrelatedness between microtubules and other cytoskeletal components. Also, it interacts with other cytoplasmic organelles, enabling binding of microtubules to the mitochondria. The N-terminal domains of tau protein enable interaction with the neuron plasma membrane.98
The MAPT gene exhibits two main haplotypes defined by a 900-kb inversion. Haplotype H1, which is considered the non-inverted sequence, is the most common in human populations. Haplotype H2, defined by the inverted sequence, is rare in African populations (roughly 6%), nearly absent in East Asians (<1%), but quite frequent in Europeans (approximately 20%), which suggests a recent history of positive selection.99 One H1 subhaplotype, H1c, has been associated with increased risk of AD in a single case-control study of autopsy-confirmed AD patients.100 Another study identified a G > A transition at rs242557 that leads to increased expression of MAPT.101 Despite these reports, the current literature and recent GWAS discourage the hypothesis of a direct association between MAPT polymorphisms and AD. However, their role in AD cannot be ruled out in light of the pathophysiological interactions that may exist between hyperphosphorylation and etiological factors for the formation of neurofibrillary tangles with concomitant cell death.
THE CLU GENE
Clusterin, or apolipoprotein J (APOJ), is a protein encoded by the CLU gene (located on chromosome 8p21–p12) and highly expressed in the liver, ovaries, testes, and brain.102 It is considered something of an enigmatic glycoprotein because of its near-ubiquitous tissue distribution and its apparent involvement in myriad biological processes, ranging from mammary gland involution to lipid transport and from hormonal biosynthesis to the pathophysiology of neurodegenerative diseases.13,102
From a biological standpoint, clusterin appears to play an important role in the pathogenesis of AD; comparison of brain tissue specimens obtained from patients with AD and normal controls shows high levels of CLU in areas affected by amyloid plaques.102 Increased clusterin expression has also been found in diseases characterized by death of large numbers of abnormal cells, such as retinitis pigmentosa, glomerulonephritis, atherosclerosis and acute myocardial infarction, or massive cell proliferation, such as gliomas and prostate cancer.102 Clusterin is abundantly expressed in neoplastic cells and appears to promote cell growth.103,104
The role of clusterin in cell cycle regulation is of interest to understanding the pathogenesis of AD, as dysfunctions in the process have been associated with AD.105,106 Higher brain functions are based on a dynamic organization of neural networks. In order to establish and continually restructure synaptic connections as part of a structural adaptation process, neurons must abstain from following the usual cell cycle and set aside the molecular mechanisms normally used in cell cycle control as an alternative pathway to ensure synaptic plasticity.105 The very presence of this alternative pathway may place neurons at risk of mistakenly converting signals received from synaptic stimulation into mitogenic activation signals, which may, in turn, lead to cell death.105
Although the CLU gene has been strongly associated with AD in recent GWAS,13,16,107 this association was first described over a decade prior to implementation of GWAS.108 Modern studies have identified the rs11136000 SNP as the source of the most significant association,13 the association between a three-SNP haplotype (rs2279590, rs11136000, rs9331888 – CCG) and AD notwithstanding.16 However, no functional studies are available to elucidate physiological effects of each of the aforementioned variants.
THE CALM/PICALM GENE
PICALM, also known as CALM (clathrin assembly lymphoid myeloid leukemia), is encoded in humans by a gene located at 11q14-21. The PICALM protein is expressed in all tissues of the human body, with the highest detectable expression in neurons, where it is distributed in the structures of the synaptic clefts109
The brains of AD patients feature a reduced number of synapses, and biochemical analyses have shown that this reduction in synaptic density is correlated with cognitive impairment.110 A more recent study suggests that synapses in the brains of AD patients may be dysfunctional well before the neurodegenerative process is established.111 PICALM may be involved in these dysfunctions, in light of the suggestion that increased expression of the protein may be implicated in the occurrence of late-onset AD.112 Therefore, it is reasonable to assume that changes in PICALM expression levels can lead to disturbances in acetylcholine-mediated synaptic transmission, thus increasing the risk of AD. An alternative hypothesis posits that PICALM might influence AD risk by affecting APP processing through endocytic pathways, thus leading to changes in Aβ levels.113
Despite their significant association, the influence of these variants on expression of PICALM and production of Aβ peptides has yet to be elucidated. PICALM gene variants have been associated with late-onset AD in independent studies.13,16,107 These studies also found that allele A of SNP rs3851179 (A > G) confers a protective effect. Subjects with this allele were at a slightly lower risk of developing AD, as was later confirmed in a replication study conducted in a European population,114 but a similar Chinese study failed to find any such benefit.
THE CR1/CD35 GENE
The CR1/CD35 gene is located on chromosome 1q32 and encodes CR1 as a single-chain transmembrane glycoprotein, with a short cytoplasmic tail and a molecular weight of approximately 200 kDa.115 It is a multifunctional, polymorphic protein expressed in the plasma membrane of erythrocytes, eosinophils, monocytes, macrophages, and B lymphocytes.
CR1 functions as a receptor for C3b and C4b and as a regulator of activation of the complement system, one of the key components of the innate immune system.116 The versatile nature of CR1 is due to the presence of several functional domains.117 The liver is the main source of complement proteins, but these peptides are also produced in other tissues and organs, including the central nervous system (CNS), where their constituent elements are synthesized by a variety of cells, such as neurons, microglia, astrocytes, oligodendrocytes, and endothelial cells.118
Recent studies have described the association between AD risk and polymorphic markers of CR1. GWAS have found SNP rs3818361 (C > T) to be associated with AD,16,107 with subjects with the T allele at a 15% higher risk for late-onset AD. This risk has been confirmed both in European and Asian populations.16,119 The effects CR1, CLU, and PICALM have been replicated in postmortem studies with different populations and multiethnic samples, and these genes are candidates for ethnicity-independent modulation of the AD phenotype.107
NEURODEGENERATIVE DISEASES AND NEUROINFLAMMMATION
Neurodegenerative diseases may be associated with chronic inflammatory conditions.120 Recurrent abnormal deposition of protein aggregates – classical activators of the inflammation cascade121– glial activation, and increased cytokine production are some of the hallmarks of inflammation present in neurodegenerative conditions, both acute (such as traumatic brain injury and stroke) and chronic (such as Parkinson's disease, amyotrophic lateral sclerosis, and AD).122
The CNS is traditionally regarded as an ‘immunologically privileged site’, that is, one not susceptible to the tissue injury caused by systemic inflammatory response. However, it is now known that the CNS has a markedly active endogenous immune system coordinated by immunocompetent cells such as the microglia and astrocytes.123 Response to infection or inflammation in the brain is distinct to the response mounted against comparable events in the periphery since classical symptoms of inflammation typically do not occur within the CNS. Furthermore, leukocyte recruitment, which occurs quickly in peripheral structures, is modest and delayed in the nervous system. These differences are partly attributable to the presence of the blood–brain barrier, which obstructs entry to inflammatory cells, pathogens and macromolecules, thus protecting the sensitive neural system from the damage typically associated with inflammation.124
Neuroinflammation in AD encompasses a wide range of complex cellular responses, such as microglial and astrocytic activation. When activated, the glia releases a variety of pro-inflammatory mediators, which may potentially contribute to neuronal dysfunction and eventually lead to cell death. These mediators trigger and feed a vicious circle that may play an essential role in the progression of AD.125 Events such as cytokine, complement protein, acute phase protein, and reactive oxygen species release, in addition to several related molecular processes,124 come together to build the current dual perception of inflammation in the brain: it plays a neuroprotective role and has deleterious effects on neuronal function. These reactions may be beneficial as limiters of ‘acute phase reactions’, but in chronic conditions, may lead to progressive neurodegeneration.126 Despite extraordinary advances concerning the inflammatory response in AD, controversy remains as to whether there is differential activation of innate and acquired immune system components on the CNS, with particular emphasis on fluctuations of immune signaling factors due to allelic variability in these mediators.127–129 Additionally, controversy exists as to whether additional negative effects can arise from these variations in the development of AD.130
Because of their diverse etiologies, basically no polymorphisms appear to be particularly involved in the pathophysiology of early-onset AD, in contrast to the current model of late-onset AD. However, the low heritability of complex diseases means that identifying and establishing relationships that can boost data on several loci to explain the influence of genetics on these conditions is a major challenge. In light of these hurdles, recent studies assessed methods meant to detect the effects of epistasis and allelic interaction between SNPs; these provide a more interactive view of multiple loci on AD and more realistic means for predicting the relative risk of AD in given genotype and/or haplotype contexts.57,131 An extensive review by Combarros et al. sought to assess the synergistic effects provided by candidate loci for protection against or risk of AD and revealed the need for well-designed further studies because of the technical and statistical limitations of current studies meant to identify epistatic effects.57 Because of its strong association with AD, the APOE gene is still the primary target of allelic interaction studies, particularly those focusing on genes related to cholesterol metabolism, oxidative stress, inflammatory pathways and other neurobiochemical networks.
Although several new loci were associated with late-onset AD in GWAS, these studies have failed to show association with important genes previously related with early-onset familial AD, such as PSEN1, PSEN2 and APP, suggesting genetic and non-genetic limitations in establishing clear cut causality relations. Also, the fact that a cast of SNPs can be significantly implicated by one study but not another may be evidence of the complexity of illustrating the interplay between single nucleotide variations and intricate traits, which reinforces the importance of replication, validation and follow-up studies.132,133 These issues are restrained, as any genetic association studies usually are, by aspects of the study design such as different genotyping platforms, number of SNPs, sample size, heterogeneity and ethnicity, type of study (case control or family based), complexity of phenotypes, quality control of experimental/clinical procedures, and statistical power. These assortments are reflected in the outcomes of each study, which, for these various reasons, should not be taken as deterministic but rather as indicative. As further GWAS are conducted, results should help elucidate the pathogenesis of AD by investigating the effects of the multitude of interactions between common variants, between rare variants, and among both groups.132
Despite their differences, the competing common disease-common variant and common disease-rare variant hypotheses may come to work in tandem like the etiology of common/classic and emergent/novel chronic illnesses involving complex interactions between a variety of frequent and rare allelic variants, lifestyle factors and socioeconomic burden.23,57,131–134 In order to achieve this understanding, exome sequencing and genome resequencing of candidate regions should be considered as a means of uncovering rare functional variants that may be in linkage disequilibrium with common SNPs. Additionally, screening for de novo or common deletions, duplications and copy number variation may shed light on certain characteristics of the pathophysiology related to the effects of gene dosage and regulatory elements.14,51 Nevertheless, omics studies (e.g. genomics, proteomics, metabolomics, transcriptomics) may have to be reconciled with modern epigenetics to enable further functional genome-wide studies of AD.
In light of the results provided by GWAS, a landmark in genetic epidemiology, future studies will be held to a higher standard of scientific rigour. We believe AD will provide an unparalleled study model for investigating genomic interactions and multigenic disease, pushing the boundaries of current knowledge on the complex phenotypes of human mental disorders.