An understanding of the environmental and genetic risk of Alzheimer’s Disease is vital for the production of more effective therapies. A recent study, published in Science, highlights the importance of studying genetic variants in the right cells and pathological context to understand more about this devastating disease.
Alzheimer’s Disease
The number of people diagnosed with dementia now slightly exceeds the number of people with cancer, and is likely to increase in future decades. However, dementia is an umbrella term, encompassing the end symptoms of a range of brain diseases, including Alzheimer’s Disease (AD).
AD progresses slowly and is characterised by the accumulation of specific proteins in the brain. Symptoms often manifest late in the disease progression, which can confound study results and impede treatment efforts. AD is a highly heritable disease, thus genetic studies are often used to identify the molecular mechanisms of the disease. However, a lack of mechanistic understanding of AD has become a major bottle-neck in the search for new drug targets.
Heritability of Alzheimer’s Disease
Heritability is defined as the proportion of phenotypic variance caused by genetic factors. This can then be used as a population-based measure for disease risk. However, its important to note that inheritance of genetic variants does not necessarily imply disease. Furthermore, not all individuals with AD possess the same genetic variants.
The most studied causal genetic variants for early onset AD include mutants in Amyloid Precursor Protein (APP) and Presenilin 1&2 (PSEN1&2). These both affect the processing of Amyloid-β Peptide (Aβ). The accumulation of Aβ is believed to be a key upstream event in AD pathogenesis.
In late-onset dementia, which manifests after the age of 65, Apolipoprotein ε4 allele (APOE4) is the only common high-risk genetic variant. The role of APOE in brain inflammation remains poorly understood. However, the gene plays a key role in cholesterol transport and lipid homeostasis, as well as Aβ aggregation, clearance and cellular uptake. It also influences synapse number and function and TAU mediated neurodegeneration. However, many of these molecular pathways are poorly understood.
Both early- and late-onset AD are highly heritable. Thus, there are increasing efforts to produce large datasets for genome-wide association studies (GWAS). These involve the direct sequencing of full genomes and the development of novel data analysis techniques to identify other hereditable factors in AD.
From Genes to Disease Mechanisms
Over 70% of genetic variants which determine the phenotypic variation of AD are “peripheral genes.” These produce indirect effects on expression and post-translational modification of gene products. Therefore, they are often not informative of the molecular mechanisms of AD.
The goal is instead to identify core genes and analyse how peripheral genes affect their expression. Also, focusing on key genes such as APOE4, which have large effects on heredity could provide key insights. On the other hand, rare variants often have large effect sizes, and therefore may be more central to the disease mechanism. Therefore this is also a key avenue of research.
APP processing and Microglial Function
The known genetic variants that contribute to AD provide strong evidence for a major pathway centred on Aβ generation, aggregation and clearance, which operated in both the early- and late-onset disease. There is also strong evidence to support microglia responses to amyloid plaques in AD. Many common variants occur in genes expressed by microglia. Furthermore, a large portion of the genetic risk of AD seems dependent on microglial responses to amyloid plaques.
For example, the gene TREM2 is responsible for the activation of microglial cells in response to amyloid plaques. Deficiencies in TREM2 expression can lead to the formation of larger plaques causing more neuritic damage. This can also reduce the recruitment of microglia to amyloid plaques. Additionally, several other AD-associated variants have been observed in genes acting downstream of TREM2, outlining its importance in AD.
Polygenic Risk
Much of the genetic risk for AD can be explained by common variation in the genome. This can be captured by single nucleotide polymorphisms (SNPs) in GWAS. These single variants alone cannot predict AD risk, but can when combined in polygenic risk scores (PRS). These are genetic scores defined by the total number of SNP risk alleles that an individual carries. These are also weighted by their contribution to their disease, also known as the effect size.
To produce a more complete picture of AD risk, SNPs in loci that are associated with AD risk, but do not reach the threshold for genome-wide significant association, should be included. These calculations can greatly improve the predictive accuracy of these studies. However, current PRS designs involve the linear combination of SNP effect sizes. These fail to account for non-linear effects, such as epistasis or SNPxSNP interactions. Biologically, it is unlikely that the risk of AD can be calculated by simple linear sums of individual SNP risks. More complex models will produce effective insights into the genetic causes of AD.
Translating the Polygenic Risk of Alzheimer’s
The main limiting factor in the translation of genetic knowledge into drug development is the lack of strong models for AD. Double, and even triple transgenic mice models overexpression human TAU, PSEN and APP genes and their associated mutations are needed to induce amyloid plaque formation. This raises the question to what extent these cellular phenotypes mimic the human situation.
Thus, more human specific research has begun. For example, in vitro 3D and organoid cultures have been developed. These use the overexpression of the APP gene to produce AD phenotypes. Furthermore, xenograft or chimeric mouse models, in which human brain cells, derived from induced pluripotent stem cells (iPSCs), are transplanted into mouse brains. The rodent brain acts as a superior physiological model for human cells compared to more artificial environments. Additionally, human neurons, microglia and astroglia have been grown in rodent brains for over a year and reproduce many human features. Although there are some confounding factors, such as immune responses, when returned to the human CNS, these microglial cells were found to regain their identity and transcriptionally resemble freshly isolated human microglia from surgical samples.
Conclusions
There has been tremendous progress made in Alzheimer’s Disease research. We understand that Alzheimer’s has a significantly large genetic risk and can now build on this knowledge to develop more accurate and complex models which represent specific mechanisms underlying Alzheimer’s disease. Deeper clinical phenotyping and biomarker identification is required to better interpret the role of genetic variation in this disease.
Ultimately, the functional insights provided by these studies will lead to more accurate diagnostics, stratification of patients for clinical trials and improved personalised medicines based on genetic profiles
