Global populations are aging, with age-related diseases, such as dementia, becoming increasingly prevalent. One of the most common but genetically complex of these is Alzheimer’s disease (AD) – a progressive disease, which starts with mild memory loss, and leads to cognitive impairment and neurodegeneration, with loss of bodily functions and death occurring in latter stages. Two major forms exist: one with an early onset age (typically <65) showing familial linking (EOAD), often through mendelian transmission; and another with a later onset age (well above 65 years) displaying no such linkage (LOAD) (Bertram and Tanzi., 2009). As genetics play a major role in the pathogenesis of AD, genome wide associated studies (GWAS) have become a key tool used in understanding the aetiology and identifying both common and rare susceptibility loci, with implications on diagnostic accuracy and potential treatments. Studies of this type test a large number of genetic markers simultaneously in a ‘hypothesis-free’ fashion, often using single nucleotide polymorphisms (SNPs) as markers which are used to cover the range of variation in the human genome.
AD pathogenesis is commonly characterized by extracellular amyloid- (A) deposits and intracellular hyperphosphorylated tau (ptau181) forming neurofibrillary tangles (See figure 1.). Both of these pathologies disrupt neuronal communication, with A oligomers binding to neuronal surface receptors and changing synaptic structure, and ptau181 leading to microtubule disintegration and cellular collapse. This progressive atrophy and subsequent loss of synaptic structure in the cerebral cortex and subcortical regions cause the cognitive deterioration associated with the condition as the patient ages (Tiraboschi et al., 2004). Mechanisms which underlie these manifestations have been scrutinised through GWAS, with several susceptibility loci being identified which highlight key proteins and druggable targets (Sloane et al., 2002). The APOE locus on chromosome 19q13.2 was of initial interest to late onset Alzheimer’s Disease (LOAD), with findings of many GWAS corroborating its significance.
The APOE gene product, Apolipoprotein E, is a human lipid binding protein, with three common isoforms expressed by 3 alleles: 2, 3 and 4, which differ from one another at 2 sites in the amino acid sequence. APOE is crucial for mediating synaptic plasticity and may be vital in cellular A metabolism through influencing clearance, synthesis and aggregation. APOE4 shows reduced efficiency in clearance, with mouse studies indicating that clearance is redirected from the usual LRP1 efflux transporter, to the VLDL receptor, which has a slower endocytotic rate due to a slower interaction with the A-apoE complex (Li et al., 2001). This redirection of metabolism leads to A deposits forming (Zhong and Weisgraber, 2009) whereas the APOE2 allele is thought to have an opposite effect, offering protection from AD, even though the mechanism is not well understood. It has been found that APOE 4 is associated with a twofold increase in susceptibility to LOAD if heterozygous, and a fivefold increase in susceptibility if homozygous, with each additional allele lowering the age of onset by 6-7 years (Breitner et al., 1999). Increased understanding of the APOE susceptibility locus has been utilised in risk prediction algorithms, which use a global genetic risk score (GRS) through combining tests for multiple susceptibility loci, allowing predictions to be made into the acceleration from mild cognitive impairment to full symptomatic AD. Such predictive tests are used in conjunction with drug trials such as the TOMMORROW clinical trial, which reviews the effects of pioglitazone in delaying the early cognitive symptoms of AD (Welsh-Bohmer et al., 2014). APOE has displayed potential in formulating novel treatments, with findings indicating that treatment outcome is often influenced by the specific APOE genotype of the individual (Crentsil, 2004). This is seen specifically within the phase II trials of bapineuzumab, with APOE 4 status stratifying individuals. It is seen that individuals carrying the 4 allele show worse AD pathology and those which are 4 negative show increased drug response (Salloway et al., 2009). Other studies have proposed treatments which target APOE gene regulation and therefore affect APOE expression, with drugs including statins showing promise in trials already.
Possible Physiological Pathway Gene Population attributable fraction for SNPs (%) Function
Immune response CR1 3.5 Regulation of complement activation in A clearance
INPP5D 3.8 Gene regulation and post translational modification of proteins, also microglial and myeloid function
MEF2C 2.8 Synaptic plasticity
TREM2 0.4 Trigger receptor involved in microglial function and regulation of immune response
Immune response/Lipid metabolism CLU 5.1 Chaperone function and cell proliferation regulation
ABCA7 2.8 Phospholipid efflux and phagocytosis
Lipid metabolism SORL1 0.91 Vesicle trafficking
APOE 30.8 Mediates synaptic plasticity and crucial in clearance, synthesis and aggregation of A
Immune response/Synaptic functioning CD33 1.8 Cell-cell interactions and cell functions in the innate and adaptive immune systems
EPHA1 3.3 Brain development, axon guidance and synapse development and plasticity
Synaptic functioning PICALM 4.5 Trafficking of synaptic vesicle proteins
CD2AP 2.6 Cytoskeletal organization and vesicle movement
PTK2B 3.6 Induction of long term potentiation in hippocampal CA1 neurones
BIN1 8.2 Clathrin-mediated endocytosis
Cytoskeletal function and axonal transport CASS4 1.0 Scaffolding protein of unknown function
FERMT2 1.2 Actin assembly and modulation of cell shape
Following the APOE locus, further GWAS were carried out and results identified susceptibility loci which were grouped into three main physiological pathways: immune response, cholesterol metabolism, and synaptic function (as illustrated in table 1). A meta-analysis was undertaken by the IGAP (International Genomics of Alzheimer’s Project) in 2013, which combined the results of the four largest GWAS, leading to a combined dataset of 54,000 individuals. The results lead to further identification of 11 novel genes, involved in cytoskeletal function, axonal transport, tau protein metabolism and cell adhesion. Finding risk loci within these pathways emphasises their potential involvement in AD pathogenesis and offers useful insights into the aetiology of the disease, providing new targets for therapy and areas for future research.
The immune pathway is an area of specific interest, due to GWAS results implicating many associated susceptibility loci, and research has been carried out into microglial and astrocyte cell function within AD pathogenesis. Microglia are resident phagocytes of the central nervous system which modulate development of circuits and play a crucial role in synaptic plasticity through neuronal circuit remodelling. Within the brains of AD sufferers, microglia are seen to cluster around neuritic plaques, whilst losing phagocytic activity and potentially becoming toxic. Rare missense mutations in genes involved in microglial function have become of specific interest to future research, exemplified by TREM2, which codes for the trigger receptor expressed on myeloid cells 2. Studies have elucidated that TREM2 leads to downregulation of A-induced microglial function and leads to a dysregulation of the cells responses to inflammation (Hickman and El Khoury, 2014). Conversely, CD33 expression is upregulated within AD microglia due to SNPs producing novel splice variants, leading to a gene product – a transmembrane protein, which has the potential to promote A42 accumulation and increase risk of AD. Determination of these loci provides new drug targets, an example being the repurposing of CD33 antibody lintuzumab (previously used for AML treatment), which has the potential to inhibit CD33 through downregulation of surface expression, reducing its effects on A aggregation. Furthermore, a greater understanding of neuroinflamation in pathogenicity could lead to a new combination of microglial and neuronal models when investigating new treatments, as opposed to the neuronal model currently used, in order to provide a better estimation of the effect of gene variants within their relevant biological system.
Establishing such genotype-phenotype correlations has led to a number of GWAS which investigate linkage of endophenotypes to specific gene variants. Benefits of these studies include the improved statistical power, enabling variants to be identified which are unable to pass the multiple test correction seen in case controlled studies (Deming et al., 2017), and provide important information regarding the rate of onset and progression of the disease. Such GWAS of endophenotypes have been performed, specifically on A42 (42 amino acid peptide form of A) and ptau181, both identifying signals located outside of the APOE region. One area of interest was within the SERPINB1 gene, with these genetic variants displaying lower levels of A42 with increased SERPINB1 expression, reinforcing conclusions that the immune response pathway has a potential role in the A cascade of AD neuropathology (Van Eldik et al., 2016). These conclusions are further corroborated by GWAS results of CSF Clusterin (CLU) levels – which show elevated levels of expression in affected brain regions of AD patients.
Clusterin (also known as apolipoprotein J/apolipoprotein15) is a pleiotropic chaperone involved in lipid transport and inflammation, which also directly influences A aggregation through endocytosis, with SNPs within CLU leading to alternative mRNA splicing and protein malfunction (Szymanski et al., 2011). One novel SNP located inside an intron has been specifically implicated and replicated across a number of studies, indicating an independent contribution in the progression from mild cognitive impairment to AD. Following locus identification, clusterin has been implicated as a potential AD biomarker, with the rate of AD progression being linked with clusterin plasma concentration. This allows CLU gene expression to be used to help predict AD progression. Predictions are based on associations made between presence of CLU allele and speed and severity of disease progression, along with hippocampal deterioration and atrophy of the entorhinal cortex – areas of the brain both associated with the early stages of disease pathogenesis. Not only does CLU provide opportunities for diagnostic test development, but by implicating CLU in modulating progression from mild cognitive impairment to symptomatic AD, novel therapeuticatargets have been identified which could aid in development of future treatments.
A main issue with developing new AD treatments is that disease pathogenesis starts well before symptoms present (Tarawneh and Holtzman, 2012). Polygenic scores (PGS) have been developed in response to this, which utilise 9 susceptibility loci from GWAS in order to predict the progression from mild cognitive impairment to AD (Lacour et al., 2016). These tests can be made more powerful if demographic, cardiovascular and lifestyle data are also incorporated, when combined with techniques such as whole exome/genome sequencing to increase accuracy of predictions. Currently, due to the lack of clinical treatments which prevent or slow the progression of AD, the tests are only useful to patients for future planning, in attempt to lessen the socio-economic impact of the disease to the family.
The elucidation of susceptibility loci through GWAS allows targetted approaches at a single gene level, with studies into gene silencing and CRISPR/CAS9 mediated genetic engineering being carried out on large animals to form models for a wide range of neurodegenerative diseases (Tu et al., 2015). The ability of the CRSIPR/CAS9 system to silence specific genes by inserting ‘loss-of-function’ mutations into the genes of interest will allowaincreased understanding of pathways involved and the effects of these genes in AD. There is also potential for CRISPR/CAS9 derived therapy, with somatic cell engineering allowing correction of some genetic mutations which are identified as significant through GWAS, and further isolated and sequenced using Next Gen Sequencing techniques (Khan et al., 2016).
Whilst genetic linkage studies identified causative variants of APP, PSEN1 and PSEN2 within the -secretase pathway, associated with familial EOAD, identification of variants in LOAD has been more difficult due to lack of clear mendelian linkage. GWAS is a revolutionary technology which has been instrumental in identifying many loci which influence LOAD susceptibility and disease progression, thereby increasing the understanding of many pathways involved in disease pathology. By identifying significant loci, predictive tests can be developed which allow risk individuals to be identified, with possibility of early diagnoses in the future. Furthermore, proteins and genes in pathways identified through GWAS are ideal targets for therapy, either by reducing expression of genes on a genetic level or regulating protein activity in order to inhibit or slow the onset of AD pathogenesis. Whilst there are no current treatments available, breakthroughs in developing a treatment involving gene silencing, as recently exemplified in Huntington’s disease, show promise for the future of all neurodegenerative diseases.