Path: mechanisms/genetics-alzheimers
Title: Genetics of Alzheimer's Disease
Tags: section:mechanisms, kind:pathology, topic:genetics, topic:gwas, topic:familial-ad, topic:sporadic-ad
Alzheimer's disease (Alzheimer's disease) is the most common cause of dementia worldwide, affecting over 55 million people globally. The genetics of Alzheimer's disease is complex, involving both rare, highly penetrant mutations that cause familial early-onset forms and common genetic variants that influence risk for the more prevalent late-onset sporadic form. Understanding the genetic architecture of Alzheimer's disease has been fundamental to elucidating disease mechanisms and developing potential therapeutic targets[@ballard2011].
Approximately 1-5% of Alzheimer's disease cases are early-onset familial forms caused by autosomal dominant mutations in one of three genes: amyloid precursor protein (APP), presenilin 1 (PSEN1), and presenilin 2 (PSEN2)[@ryman2018]. These mutations converge on the amyloidogenic pathway, leading to increased production or aggregation of amyloid-beta (amyloid-beta) peptides. The remaining cases are classified as late-onset Alzheimer's disease (LOAD), which has an estimated heritability of 60-80% and arises from a complex interplay of multiple genetic variants and environmental factors[@jansen2019].
The advent of genome-wide association studies (GWAS) has identified over 40 genomic loci associated with Alzheimer's disease risk, revealing pathways involved in immune response, lipid metabolism, synaptic function, and endosomal trafficking[@kunkle2019][@lambert2013]. The apolipoprotein E (APOE) ε4 allele remains the strongest known genetic risk factor for LOAD, while recent studies have highlighted the importance of rare coding variants in genes such as TREM2, ABCA7, and PLCG2[@jonsson2013][@sims2017]. This expanding genetic landscape provides insights into disease pathogenesis and opportunities for precision medicine approaches.
The amyloid precursor protein gene located on chromosome 21q21 is fundamental to Alzheimer's disease pathogenesis as the source of amyloid-beta peptides. Over 50 pathogenic mutations in APP have been identified, predominantly causing early-onset familial Alzheimer's disease with complete penetrance[@van2016]. These mutations cluster around the β-secretase and γ-secretase cleavage sites, promoting amyloidogenic processing.
The Swedish double mutation (APP KM670/671NL) located at the β-secretase cleavage site was the first Alzheimer's disease-causing mutation identified and dramatically increases amyloid-beta production by enhancing β-secretase cleavage[@mullan1992]. The London mutation (APP V717I) at the γ-secretase site shifts amyloid-beta production toward the longer, more aggregation-prone Aβ42 species. The Flemish mutation (APP A692G) results in both cerebral amyloid angiopathy and Alzheimer's disease, illustrating the phenotypic variability of APP mutations.
Duplication of the APP locus causes autosomal dominant Alzheimer's disease in families, particularly in those with early-onset disease and cerebral amyloid angiopathy, providing definitive evidence that APP overexpression is sufficient to cause Alzheimer's disease[@sleegers2006]. Additionally, the APP A673T variant, found in the Icelandic population, provides protection against Alzheimer's disease and cognitive decline, representing a natural experiment supporting amyloid-targeted therapeutic approaches[@jonsson2012].
PSEN1, located on chromosome 14q24.3, encodes the catalytic subunit of the γ-secretase complex. Over 300 pathogenic mutations in PSEN1 have been identified, making it the most common cause of familial early-onset Alzheimer's disease, accounting for approximately 50-70% of autosomal dominant cases[@lanoo1994]. PSEN1 mutations have the earliest age of onset, typically between 30-50 years.
Most PSEN1 mutations are missense substitutions that alter γ-secretase activity, leading to increased production of the aggregation-prone Aβ42 species relative to Aβ40[@scheuner1996]. This shift in the Aβ42/Aβ40 ratio is believed to be central to pathogenesis, though some mutations may also affect Notch signaling and other γ-secretase substrates. Notably, certain PSEN1 mutations, such as PSEN1 ΔE9, cause variant phenotypes including spastic paraparesis, which correlates with intensive Aβ42 deposition in cerebral vessels.
The identification of PSEN1 mutations demonstrated that disruption of γ-secretase activity is sufficient to cause Alzheimer's disease, highlighting the importance of the amyloidogenic pathway. Studies of PSEN1 knock-in mice have confirmed that mutant PSEN1 increases Aβ42 levels and accelerates amyloid plaque formation, providing mechanistic validation[@guijarro2010].
PSEN2 on chromosome 1q42.13 encodes presenilin 2, another catalytic component of the γ-secretase complex with high homology to PSEN1. Approximately 40 pathogenic mutations in PSEN2 have been associated with familial Alzheimer's disease, though some variants have incomplete penetrance or uncertain pathogenicity[@ryman2021].
PSEN2 mutations generally cause later onset than PSEN1 (median age 50-65 years), but phenotypic variability is extensive. The PSEN2 N141I mutation, identified in Volga German families, is one of the most extensively studied pathogenic variants and demonstrates nearly complete penetrance with mean onset at 52 years[@nochlin1999]. Interestingly, some PSEN2 variants, including the missense mutation PSEN2 R62H, have been identified in apparently asymptomatic elderly individuals, suggesting possible reduced penetrance or protective effects.
Functional studies have demonstrated that most PSEN2 mutations shift the Aβ42/Aβ40 ratio upward, similar to PSEN1 mutations. However, PSEN2 may have additional γ-secretase-independent functions, and its role in cellular processes such as calcium homeostasis and autophagy has been implicated in neurodegeneration[@lee2010].
The APOE gene on chromosome 19q13.32 encodes apolipoprotein E, a polymorphic protein involved in lipid transport and neuronal repair. APOE has three common alleles (ε2, ε3, ε4) resulting from amino acid substitutions at positions 112 and 158[@mahley2000]. The ε4 allele is the most significant genetic risk factor for late-onset Alzheimer's disease, while the ε2 allele appears to be protective.
APOE ε4 carriers have a 3-4 fold increased risk of Alzheimer's disease compared to non-carriers, with heterozygotes having approximately 50% increased risk and homozygotes having 10-12 fold increased risk[@strittmatter1996]. The effect is dose-dependent, with earlier age of onset in ε4/ε4 homozygotes (mean onset approximately 68 years) compared to ε3/ε3 carriers (mean onset approximately 75 years). Population studies suggest that approximately 40-65% of Alzheimer's disease patients carry at least one ε4 allele.
Multiple mechanisms have been proposed for APOE ε4-mediated risk, including: increased amyloid-beta aggregation and reduced clearance; altered lipid metabolism affecting neuronal health; impaired synaptic plasticity; neuroinflammation through differential microglial activation; and reduced cerebral glucose metabolism[@huang2014]. The APOE ε2 allele appears to have opposing effects, promoting amyloid-beta clearance and reducing aggregation, consistent with its protective effect.
Triggering receptor expressed on myeloid cells 2 (TREM2) is a cell surface receptor on microglia that plays a critical role in immune response and lipid metabolism. Rare coding variants in TREM2 were first associated with Alzheimer's disease risk in 2013, representing a major breakthrough in understanding the role of neuroinflammation in Alzheimer's disease pathogenesis[@guerreiro2013].
The R47H variant (rs75932628) confers approximately 3-fold increased Alzheimer's disease risk, similar to the effect of one APOE ε4 allele. This variant impairs microglial function, including the ability to phagocytose amyloid-beta plaques and support neuronal survival[@ulrich2017]. Additional risk variants include R62H and D87N, while the R47H variant has been consistently replicated across multiple populations.
Homozygous TREM2 loss-of-function variants cause Nasu-Hakola disease, a rare disorder characterized by bone cysts and early-onset dementia, demonstrating that complete loss of TREM2 function is sufficient to cause neurodegeneration[@paloneva2001]. Studies in mouse models have shown that TREM2 deficiency reduces microglial clustering around amyloid plaques, suggesting a protective role for microglia in containing amyloid pathology.
Large-scale GWAS have identified numerous additional Alzheimer's disease risk loci, revealing biological pathways not immediately apparent from early genetic discoveries. The clusterin (CLU) gene on chromosome 8p12 encodes a complement inhibitor involved in amyloid-beta clearance, with common variants conferring approximately 10-15% increased risk[@harold2009].
The ABCA7 gene on chromosome 9p24 encodes an ATP-binding cassette transporter involved in lipid homeostasis and phagocytosis. Loss-of-function variants, particularly in African American populations, significantly increase Alzheimer's disease risk, highlighting the importance of microglial lipid metabolism in disease[@reitz2015].
Other significant GWAS hits include: CR1 (complement receptor 1), involved in complement activation and microglial clearance; PICALM (phosphatidylinositol binding clathrin assembly protein), involved in endocytic trafficking; MS4A4A/MS4A6E, involved in calcium signaling; CD2AP (CD2-associated protein), involved in synaptic function; and EPHA1, involved in cellular repulsion[@lambert2010]. The cumulative effect of these common variants, each contributing modest risk, shapes the polygenic architecture of late-onset Alzheimer's disease.
Pathogenic mutations causing autosomal dominant Alzheimer's disease share a common mechanistic theme: they all increase the production or aggregation of amyloid-beta peptides. Understanding these mutations has been fundamental to developing the amyloid hypothesis of Alzheimer's disease pathogenesis.
APP mutations can be classified based on their location within the gene and their effect on amyloid-beta processing. Mutations at the β-secretase cleavage site (Swedish mutations) increase overall amyloid-beta production by enhancing BACE1 cleavage. Mutations at the γ-secretase cleavage site alter the Aβ42/Aβ40 ratio, favoring the longer, more aggregation-prone Aβ42 species. Mutations within the amyloid-beta sequence itself (such as the Arctic mutation E22G) promote amyloid-beta aggregation and plaque formation[@nilsberth2001].
Presenilin mutations function through γ-secretase dysfunction. While some mutations reduce overall γ-secretase activity, the predominant effect of most pathogenic mutations is to alter the specificity of γ-secretase cleavage, increasing the relative production of Aβ42[@xia2020]. This shift in product profile is sufficient to initiate the amyloid cascade, as demonstrated by the fact that overexpression of Aβ42 in mouse models is sufficient to cause amyloid deposition and downstream pathology.
The penetrance of pathogenic mutations in APP, PSEN1, and PSEN2 is high but not absolute. Age-related decreases in penetrance have been observed, suggesting that genetic modifiers or environmental factors can influence disease expression. For PSEN1, the median age of onset varies by mutation, ranging from under 40 years to over 60 years, indicating mutation-specific effects on disease timing[@ryman2020].
The specific mutation in a familial Alzheimer's disease gene can influence the clinical phenotype. PSEN1 mutations are associated with the earliest onset and most rapid progression. Some PSEN1 mutations, particularly those causing early-onset disease with spastic paraparesis, are associated with atypical histopathology, including cotton wool plaques[@larner2019].
APP mutations at position 692 (Flemish) and 693 (Italian, Dutch) cause more prominent cerebral amyloid angiopathy, leading to hemorrhagic strokes in addition to cognitive decline. The APP Swedish mutation causes typical Alzheimer's disease with relatively early onset. These observations suggest that while all APP and presenilin mutations ultimately lead to amyloid pathology, the precise biochemical effects can influence the pattern of neuropathology and clinical presentation.
Common genetic variants influencing Alzheimer's disease risk have been identified through genome-wide association studies, case-control studies, and more recently, genome-wide association meta-analyses involving tens of thousands of subjects.
The APOE ε4 allele represents the strongest common genetic risk factor for late-onset Alzheimer's disease. Beyond the well-established ε4 risk allele, more subtle effects of the APOE locus continue to be discovered. The APOE ε3 allele appears to be the default high-frequency allele in most populations, while the ε2 allele provides protection[@belloy2019].
Recent studies have identified additional variants within the APOE locus that modify risk independently of ε2/ε3/ε4 haplotypes. The rs429358 and rs7412 variants defining the ε2/ε3/ε4 system are in linkage disequilibrium with other variants that may influence expression or function. Studies in populations of African ancestry have identified APOE variants with distinct effects, including the ε4 duplication that confers extremely high risk[@farrer1997].
The concept of polygenic risk scores (PRS) aggregates the effects of thousands of genetic variants to predict individual risk for Alzheimer's disease. PRS developed from large GWAS datasets can identify individuals at 2-3 fold increased risk compared to the population average, though the clinical utility of PRS for Alzheimer's disease remains under investigation[@escottprice2017].
The predictive power of Alzheimer's disease PRS is limited by the small effect sizes of most risk variants. At current, PRS explain approximately 10-15% of the variance in Alzheimer's disease risk, compared to APOE alone which explains approximately 5-10%. Combined PRS including APOE and genome-wide variants offer improved prediction, but validation in diverse populations and determination of clinical thresholds remain active areas of research[@adesviader].
Alzheimer's disease exhibits a dual genetic architecture: rare mutations with high penetrance cause monogenic early-onset forms, while common variants with small effect sizes collectively contribute to polygenic late-onset risk.
Monogenic autosomal dominant Alzheimer's disease accounts for approximately 1-5% of all cases and is caused by highly penetrant mutations in APP, PSEN1, or PSEN2. These mutations are typically inherited in an autosomal dominant pattern with complete or nearly complete penetrance. The age of onset in mutation carriers is relatively predictable, though modifier genes and environmental factors can influence timing[@bird2008].
The identification of monogenic Alzheimer's disease families has been critical for understanding disease pathogenesis and developing therapeutic interventions. Preclinical studies in mutation carriers have enabled biomarker development and prevention trial design. The Dominantly Inherited Alzheimer Network (DIAN) and similar studies have established that biomarker changes begin 15-20 years before clinical onset in mutation carriers, providing a window for preventive interventions[@bateman2012].
Late-onset Alzheimer's disease is highly polygenic, with genome-wide association studies identifying over 40 independent risk loci. The heritability of LOAD is estimated at 30-60%, but common SNPs collectively explain only approximately 10-15% of variance, indicating that much heritability remains unexplained, potentially in rare variants or structural variation[@ridge2013].
The polygenic nature of LOAD means that risk is distributed across many genes with small individual effects. Pathway analyses of GWAS hits reveal enrichment for processes including: immune system regulation (CLU, CR1, TREM2); lipid metabolism (APOE, ABCA7); synaptic function (PICALM, CD2AP); and endosomal trafficking (BIN1, PTK2B). This diverse set of pathways suggests that Alzheimer's disease pathogenesis involves multiple upstream insults that converge on common downstream mechanisms[@jones2010].
The risk of Alzheimer's disease is shaped by the interaction between genetic susceptibility and environmental factors, though the molecular mechanisms underlying these interactions remain incompletely understood.
Epidemiological studies have identified modifiable risk factors for Alzheimer's disease, including cardiovascular health, education, physical activity, diet, and traumatic brain injury. How these environmental factors interact with genetic susceptibility is an area of active investigation.
The APOE genotype modifies the effect of several environmental exposures on Alzheimer's disease risk. For example, the adverse effects of smoking on Alzheimer's disease risk are stronger in APOE ε4 carriers, while the protective effect of physical activity may also be genotype-dependent[@reitz2016]. These gene-environment interactions suggest that lifestyle interventions might be particularly beneficial for those at highest genetic risk.
Environmental factors can influence gene expression through epigenetic modifications, including DNA methylation, histone modifications, and non-coding RNAs. Studies have identified epigenetic changes in Alzheimer's disease brain tissue, though distinguishing primary causal changes from downstream effects remains challenging.
DNA methylation clocks based on epigenetic patterns can predict biological age, and accelerated epigenetic aging has been associated with Alzheimer's disease risk. Studies have identified specific CpG sites where methylation is altered in Alzheimer's disease cases, though these changes often reflect disease-related processes rather than primary susceptibility factors[@bakulski2019].
Current research in Alzheimer's disease genetics focuses on several key areas: identifying additional risk variants, especially in diverse populations; understanding functional mechanisms of risk variants; developing precision medicine approaches; and translating genetic discoveries into therapeutic targets.
Most genetic studies have been conducted in populations of European ancestry, limiting generalizability. Recent studies in African American, Hispanic, and Asian populations have identified population-specific variants and demonstrated that effect sizes can differ across ancestry groups[@rajabli2023]. The APOE ε4 effect is stronger in populations of African ancestry compared to European ancestry, while other variants such as ABCA7 show more pronounced effects in African Americans.
Large-scale studies in diverse populations are essential for understanding the full genetic architecture of Alzheimer's disease and ensuring that genetic discoveries benefit all populations. The Alzheimer's Disease Genetics Consortium and similar efforts are working to increase diversity in genetic studies.
Moving from genetic associations to mechanistic understanding requires functional follow-up. Studies using induced pluripotent stem cells, mouse models, and cellular systems are investigating how risk variants affect molecular pathways. For TREM2, this work has revealed the importance of microglial function in Alzheimer's disease and led to clinical trials of TREM2-targeting antibodies[@schenk2022].
Single-cell genomics approaches are enabling cell-type-specific analysis of Alzheimer's disease risk gene expression. Studies have identified Alzheimer's disease-risk-associated expression quantitative trait loci (eQTLs) that influence gene expression in specific cell types, providing insight into cell autonomous effects of genetic risk.
Genetic discoveries are informing therapeutic development across multiple approaches. Amyloid-targeting therapies based on the understanding of APP and presenilin mutations have been developed, including monoclonal antibodies against amyloid-beta and BACE inhibitors. The failure of many of these approaches has prompted reconsideration of the amyloid hypothesis and exploration of additional targets[@panza2019].
Genetic findings have also highlighted novel therapeutic targets. TREM2 agonists are in development based on the protective effect of microglial activation. Genes involved in lipid metabolism (APOE, ABCA7) suggest that modulating brain lipid homeostasis might be beneficial. The identification of protective variants, such as APP A673T, provides proof-of-concept that interfering with amyloid production can prevent Alzheimer's disease[@oconnor2021].
The genetics of Alzheimer's disease has revealed a complex architecture spanning rare highly penetrant mutations causing familial early-onset disease and common variants with modest effects shaping late-onset risk. Discoveries in APP, PSEN1, and PSEN2 established the centrality of amyloid in Alzheimer's disease pathogenesis, while GWAS have revealed the importance of immune, lipid metabolic, and synaptic pathways. The identification of TREM2 variants highlighted neuroinflammation as a key disease mechanism. Ongoing research aims to identify additional genetic variants, understand functional mechanisms, and translate findings into precision medicine approaches and therapeutic interventions.