Protective genetic variants against Alzheimer’s disease

Abstract

Genetic studies can offer powerful insights for the development of disease-modifying therapies for Alzheimer’s disease. Protective genetic variants that delay the onset of cognitive impairment have been found in people with sporadic Alzheimer’s disease and in carriers of mutations that usually cause autosomal-dominant Alzheimer’s disease in mid-life. The study of families who carry autosomal dominant mutations provides a unique opportunity to uncover genetic modifiers of disease progression, including rare variants in genes such as APOE and RELN. Understanding how these variants confer protection can help identify the biological pathways that contribute to cognitive resilience, such as the heparan-sulphate proteoglycan–APOE receptor pathway, the TREM-2-driven signalling pathways in the microglia, and phagocytosis. Therapies able to replicate the beneficial effects of these natural defences could provide novel strategies for slowing or preventing the progression of Alzheimer’s disease.

Introduction

Alzheimer’s disease, the most common form of dementia, is characterised neuropathologically by amyloid β peptide plaques and neurofibrillary tangles.¹ Advances in understanding amyloid and tau pathology led to the development of biomarkers for presymptomatic and early diagnosis,² and to the accurate postmortem diagnosis of Alzheimer’s disease.^3-5 However, progress in treatment remains slow. Animal models have reinforced the link between toxic amyloid β and phosphorylated tau (p-tau) intermediates with cognitive decline,¹ emphasising the importance of early detection of atypical concentrations of these proteins through fluid or imaging biomarkers.⁵ Pathological amyloid β and p-tau aggregates first appear in specific cortical and subcortical areas,⁶ with tau pathology more closely linked to clinical progression of Alzheimer’s disease than amyloid β deposition.⁷

Compared with the extensive research on Alzheimer’s disease risk so far, less is known about the mechanisms underlying resistance and resilience to the disease. Two cases of autosomal-dominant Alzheimer’s disease showcase these protective phenomena.^8,9 A carrier of the PSEN1 Glu280Ala mutation who was homozygous for the APOE3 Christchurch variant (also known as Arg136Ser and abbreviated throughout as APOE3Ch) had region-specific resistance to tau pathology.^8,10 The other case was that of a PSEN1 Glu280Ala mutation carrier with a rare RELN gene variant (His3447Arg and referred to throughout as RELN-COLBOS).⁹ In both cases, symptom onset was delayed by more than two decades compared with typical PSEN1 Glu280Ala carriers from the same Colombian kindred, who usually develop mild cognitive impairment by age 44 years and dementia by age 49 years.¹¹ These cases help distinguish resistance (a term that refers to the cellular or molecular processes that prevent or reduce pathology) from resilience (a term that refers to the capacity of certain individuals for maintaining cognitive functions despite having typical Alzheimer’s disease pathology).¹²

In this Review, we describe the evidence on protective genetic variants in Alzheimer’s disease and evaluate how these findings could guide the development of novel treatment strategies aiming to slow or prevent disease progression by mimicking or boosting these natural defence mechanisms.

Protective genetic variants against sporadic Alzheimer’s disease

Genome-wide association studies have allowed the discovery of several genetic variants that are protective against sporadic Alzheimer’s disease, beyond the widely known APOE allele ε2.^13-15 The development of geneticbased metrics seems crucial for identifying and monitoring individuals at high risk of sporadic Alzheimer’s disease, and for selecting participants for clinical trials testing early-stage interventions.^16-18 Additionally, identification of resilience-associated variants can inform the development of disease-modifying therapies by uncovering biological pathways involved in Alzheimer’s disease progression (panel).^16,19

Among genetic risk factors, APOE remains the strongest and most prevalent risk factor for sporadic Alzheimer’s disease, comprising three major allelic variants: APOEε2, APOEε3, and APOEε4, with either a protective (APOEε2), physiologically neutral (APOEε3), or detrimental (APOEε4) effect on Alzheimer’s disease pathology.²⁰ In sporadic Alzheimer’s disease, the APOEε2 allele (which encodes ApoE2) is the most common protective variant, with an allele frequency estimated to be at 7% of the population worldwide, albeit with geographical variability.^21-23 People who carry APOEε2 have a 10-year delay in age of disease onset as compared with those who carry APOEε4, who have an age of onset between 63·8 years and 76·0 years for people who are homozygous for the ε4 allele, and between 66·1 and 80·5 years for those who are heterozygous (ε3/ε4).^24,25

In 4-month-old APP/PS1/APOE4 mouse models of Alzheimer’s disease pathology, the intracerebral injection of viral vectors expressing APOEε2 in the ventricles has been associated with reduced amyloid β plaque deposition in the cortex, decreased synaptic loss, and less microglial activation 2 months after injection, compared with vehicle control-treated mice.²⁶ In addition to APOEε2, other APOE variants have been linked to protective effects. Among them, two rare missense variants in APOEε3 (Val236Glu or ApoE3-Jacksonville) and APOEε4 (Arg251Gly) have been reported to substantially lower the risk of Alzheimer’s disease by more than 60% and more than 50%, respectively.^27,28 The APOEε4 allele, conversely, is the strongest genetic risk factor for Alzheimer’s disease.²⁰ APOEε4 knockout has therefore been proposed as a therapeutic strategy. This hypothesis is supported by evidence showing that loss-of-function variants on APOEε4 are associated with Alzheimer’s disease protection in human beings.²⁹ Lowered ApoE concentrations in a P301S/APOEε4 mouse model, by use of antisense oligonucleotides targeting APOE mRNA, were associated with reduced tauopathy, brain atrophy, inflammation, and synaptic loss.³⁰

Findings from genome-wide association studies suggest an association between APOE and other genes, such as TMEM106B, in influencing Alzheimer’s disease risk.³¹ Specifically, a study with human samples found that TMEM106B, APOE, and amyloid β effects converge to alter myelination and lysosomal gene expression, whereas the TMEM rs1990621 variant has been associated with increased neuronal density.^32,33 Another key gene, TREM-2, is expressed in microglia.³⁴ This gene plays a crucial role in microglia proliferation, survival, migration, and phagocytosis.³⁵ In mouse models of Alzheimer’s disease, TREM-2 haplodeficiency and deficiency can lead to reduced microglial clustering around amyloid β plaques, suggesting that TREM-2 is required for plaque-associated microglial responses.³⁶ Stimulating TREM-2 with agonistic antibodies could be a new therapeutic approach, as it can enhance microglial metabolism in several models of Alzheimer’s disease.^37,38

Additionally, ABCA7 has been highlighted by genomics, transcriptomics, and methylomics approaches for its roles in amyloid β clearance via phagocytosis and in lipid metabolism by promoting the release of phospholipids and cholesterol. Several ABCA7 variants have been highlighted as protective, such as rs72973581³¹ and Val1613Met, via a suggested gain-of-function mechanism.³⁹ Conversely, loss-of-function variants, such as rs4147929, have been associated with increased Alzheimer’s disease risk in Icelandic populations (OR 2·12) and in European and US populations (combined OR 2·03).⁴⁰ These findings highlight ABCA7 as a possible genetic biomarker for Alzheimer’s disease, warranting further investigation into its precise mechanisms⁴¹ given that some variants have been associated with increased risk for Alzheimer’s disease in specific populations, such as African American populations (rs11550680, rs14076058, rs567222111, rs10405305, and rs3764647) and Asian populations (rs3764650).⁴²

APP, which encodes the amyloid precursor protein, is central to Alzheimer’s disease pathology. Successive cleavages of the amyloid precursor protein by β secretase and γ secretase produce amyloid β peptides that can aggregate and form plaques in the brain. In addition to APP, mutations in the PSEN1 and PSEN2 genes are generally associated with autosomal dominant Alzheimer’s disease, as these genes encode the major enzymatic component of γ secretase.⁴³ In a set of wholegenome sequencing data from 1795 Icelandic individuals, the Ala673Thr mutation of the APP gene was found to be more prevalent in the healthy control population than in people with Alzheimer’s disease.⁴⁴ This mutation also dampened amyloid β toxicity in a mouse model.⁴⁵

Emerging evidence suggests that the PICALM gene modulates multiple pathways of Alzheimer’s disease pathogenesis, including the processing of the amyloid precursor protein and amyloid β transcytosis, and substantially affects the progression of tau pathology in Alzheimer’s disease models.⁴⁶ The PICALM rs3851179A single nucleotide polymorphism (SNP) has shown protective effects against Alzheimer’s disease in populations from France, Italy, Lebanon, Spain, and Türkiye.⁴⁷

Independently from previous findings on genes such as PICALM and APP, multiple genome-wide association studies have reported the rs11136000 SNP of the CLU gene, encoding for clusterin, as a protective locus for Alzheimer’s disease risk in European and US cohorts. This variant seems to upregulate clusterin concentrations in the temporal cortex of carriers.^48-50 Furthermore, a case–control association study in a cohort of Czech individuals⁵¹ replicated this protective effect, also involving reduced aggregation of amyloid β.^48,51

The expression of another gene of interest, PLCG2, is increased in several brain regions in patients with late-onset Alzheimer’s disease compared with cognitively unimpaired participants.⁵² Increased PLCG2 is also highly correlated with brain amyloid burden in both patients with late-onset Alzheimer’s disease and in Alzheimer’s disease model mice.⁵³ Specifically, the Pro522Arg variant could reduce Alzheimer’s disease progression by mitigating tau pathology in the presence of amyloid pathology by acting as a functional hypermorph, or in other words, a gain-of-function variant.^53,54 TREML2 rs3747742 and rs6915083 SNPs were associated with reduced Alzheimer’s disease risk in separate studies.^55,56 TREML2 is involved in the regulation of inflammation by regulating microglia function.⁵⁷

Finally, a study conducted in the Alzheimer’s Disease Genetics Consortium, Alzheimer’s Disease Sequencing Project, and UKB cohorts reported that two missense, potential loss-of-function variants in the fibronectin 1 (FN1) gene might protect against APOEε4-mediated Alzheimer’s disease pathology. In cognitively unimpaired individuals homozygous for APOEε4, brain expression showed statistically significantly lower FN1 deposition and reduced reactive gliosis compared with people with homozygous APOEε4 and Alzheimer’s disease. Furthermore, fibronectin loss of function in zebrafish models can enhance gliovascular remodelling at the blood–brain barrier, increase microglial activity, and reduce gliosis. These findings suggest that FN1 could act as a downstream driver of APOEε4-mediated pathology and cognitive decline in Alzheimer’s disease.⁵⁸

Protective genetic variants against autosomal dominant Alzheimer’s disease

In a Colombian woman carrying the PSEN1 Glu280Ala variant who was homozygous for the APOE3Ch variant, the onset of cognitive symptoms was delayed by more than 30 years, compared with typical carriers from the same kindred, who generally develop mild cognitive impairment around age 44 years and progress to dementia by age 49 years.⁸ In this woman, PET imaging using ¹⁸F flortaucipir as a tracer revealed an atypical pattern of tau accumulation, of reduced magnitude and confined primarily to the occipital cortex, with little involvement of medial temporal and other cortical regions. Postmortem analysis confirmed sparse tau neuropathology, with a higher concentration in posterior regions than in the frontal cortex. This analysis also showed homoeostatic astrocytes and minimal vascular pathology, even in the presence of an elevated amyloid burden.^8,59,60 Morphological analyses of her brain consistently confirmed the presence of a larger and more complex astroglia morphology and reduced vascular pathology, compared with PSEN1 Glu280Ala carriers without the APOE3Ch variant. Furthermore, the astrocyte distribution in the cortical layer II was very similar to that found in healthy brains, and low concentrations of p-tau were found in layers I-III of the frontal cortex, which possibly contributed to her cognitive resilience.⁶⁰ Heterozygous carriers of APOE3Ch and PSEN1 Glu280Ala from the same Colombian kindred, including both female and male individuals, have also been studied. These individuals showed a delay of approximately 5 years in the onset of mild cognitive impairment and 4 years in dementia onset compared with typical PSEN1 Glu280Ala carriers. A neuroimaging study in two APOE3Ch carriers revealed reduced tau deposition and relatively preserved brain metabolism compared with non-carriers, whereas postmortem analyses showed less amyloid-related vascular pathology in these two individuals than in APOE3Ch non-carriers.^60,61

Contrasting the protective effect observed in Colombian carriers, a study of a family from Spain reported a case of early-onset Alzheimer’s disease at age 53 years in a male individual who was heterozygous for the APOE3Ch variant, and without any mutation causing Alzheimer’s disease, whereas a female sibling with the same variant developed Alzheimer’s disease at age 66 years. Postmortem analysis of the male carrier showed amyloid angiopathy and Lewy body copathology consistent with advanced Alzheimer’s disease.⁶² These findings suggest that APOE3Ch heterozygosity does not universally confer protection against Alzheimer’s disease and highlight the complexity of gene–environment and gene–gene interactions.

Experimental studies have examined the mechanisms underlying the protective effects of APOE3Ch, and findings suggest that this genetic variant reduces its interaction with heparan sulfate proteoglycans.⁶³ This reduced binding of ApoE3Ch to heparan sulphate proteoglycans appears to limit tau deposits in brain regions that are typically affected in Alzheimer’s disease, such as the medial temporal lobe.^8,59 By interfering with this pathway, ApoE3Ch might help modulate downstream processes that drive neurodegeneration.

Neurons obtained from induced human and mouse pluripotent stem cells carrying human APOE4 and either the homozygous or heterozygous Christchurch variant have also uncovered some potential mechanisms of neuroprotection.^64-66 Specifically, given that APOE4 is a established major risk factor for sporadic Alzheimer’s disease,²⁰ establishing whether the Christchurch variant could also protect in the presence of the APOE4 allele was crucial. Findings have confirmed that the homozygous APOE4Ch variant provided protection against tau pathology, neurodegeneration, and neuroinflammation, whereas heterozygotes showed a partial protection against neurodegeneration and neuroinflammation, but not against tau pathology.⁶⁵ A novel antibody designed to bind to the heparin-binding domain of ApoE3 in the R136 region, the 7C11 antibody, has been tested in both APOE4 knock-in and MAPT^P301S transgenic murine models, showing reduced tau phosphorylation (p-tau S396) in the brains of systemically treated APOE4 knock-in mice,⁶⁷ and reduced tau pathology in the retina of MAPT^P301S mice. In murine models carrying APOE3Ch, positive effects on amyloid pathology, amyloid-induced tauopathy, and tau-driven neurodegeneration have been shown.⁶⁸ Additionally, improvements in microglial response and enhanced myeloid cell phagocytosis have been observed.⁶⁴ Despite these findings, the optimal balance between microglial activation and suppression remains uncertain.

Overall, inhibiting the interaction between ApoE and heparan sulphate proteoglycans might serve as a promising therapeutic strategy that could affect tau aggregation and spreading. Additionally, this inhibition might influence the severity, progression, and clinical presentation of autosomal-dominant Alzheimer’s disease by reducing ApoE-mediated neurotoxicity and tau hyperphosphorylation. Findings on APOE3Ch also support that the interaction between ApoE and heparan sulphate proteoglycans could serve as a promising therapeutic target for Alzheimer’s disease.⁶⁷

A resilient male individual from the Colombian PSEN1 Glu280Ala kindred who remained cognitively unimpaired until his late 60s (more than 20 years after the expected age of dementia onset) was found to be a carrier of the rare RELN-COLBOS variant.⁹ In vitro and in vivo molecular genetic studies found that Reelin-COLBOS statistically significantly increased the phosphorylation concentrations of phosphorylated disabled homolog 1 (DAB1) in neurons, possibly via the higher binding affinity of Reelin-COLBOS to heparan sulphate proteoglycans, thus favouring increased local concentrations of Reelin in the vicinity of the APOE receptors. To further validate the protective mechanism of RELN-COLBOS, a knock-in mouse model carrying the RELN-COLBOS variant homologous to the human variant confirmed a gain-of-function effect via increased phosphorylated DAB1 concentrations; this effect was associated with reduced tau pathology and preserved cognitive performance when these mice were crossed with a transgenic model of tau pathology. Interestingly, the mouse model also indicated a sexually dimorphic effect of the RELN-COLBOS variant, with statistically significantly increased phosphorylated DAB1 concentrations only in male mice.

Postmortem examination of the resilient Colombian male carrier of the RELN-COLBOS variant showed severe Alzheimer’s disease pathology despite having high neuronal density and low tau pathology in the entorhinal cortex, compared with other individuals from the same kindred. These findings further support the protective role of this RELN-COLBOS variant.⁹ The unique characteristics of this example provide valuable insights, complemented by in vitro and animal studies, and support the role of Reelin in Alzheimer’s disease resilience and as a potential therapeutic target that could be used to mimic the effect of RELN-COLBOS.⁹ Physiologically, the glycoprotein Reelin, encoded by the RELN gene, has several important functions in the brain, such as regulating neuronal migration, dendritic growth and branching, dendritic spine formation, synaptogenesis, and synaptic plasticity.⁶⁹ Reelin promotes long-term potentiation in the hippocampus by activating ApoEr2 and VLDL receptors and their downstream signalling via Dab1, which strengthens NMDA receptor function.⁷⁰ Experimental models show that reduced Reelin signalling can worsen age-related learning and memory deficits when combined with Alzheimer’s disease-related tau mutations.⁷¹ Additionally, Reelin mitigates amyloid β toxicity and tau hyperphosphorylation, two hallmarks of Alzheimer’s disease, supporting its role in preserving cognitive function with ageing.⁷⁰ Other evidence suggests that Reelin signalling might contribute to cognitive resilience and potentially lower dementia risk.^69-71 Studies in mice have shown that the RELN gene provides protection against amyloid β toxicity, compared with the low RELN expression in conditional knock-out mouse models.⁷² A study reporting a single-cell transcriptomics atlas identified vulnerable excitatory and inhibitory neurons that are depleted in specific brain regions in Alzheimer’s disease, providing evidence that the Reelin signalling pathway influences their vulnerability.⁷³

Lowering Reelin in entorhinal cortex layer II neurons through microRNA-mediated knock down reduces intracellular amyloid β without affecting amyloid precursor proteins, which suggests a direct link between Reelin and intracellular amyloid β accumulation.⁷⁴ Impaired Reelin signalling through amyloid β peptide binding to Reelin, alterations in Reelin processing and glycosylation, or reduced formation of active Reelin homodimers correlates statistically with higher tau phosphorylation in cultures of primary neurons treated with amyloid β, in the lateral entorhinal cortex of rats aged 24 months with cognitive impairment, and in various mouse models of tauopathy.⁷⁵ Furthermore, Reelin overexpression can improve memory and counter tau pathology in transgenic mouse models expressing human tau.⁷⁶ Patients with Alzheimer’s disease have distinct Reelin proteolytic fragments in CSF, including an approximately 500 kDa Reelin species, and increased expression of genes of the Reelin pathway, highlighting its potential role in Alzheimer’s disease.⁷⁷

A longitudinal analysis of a cohort of individuals carrying the PSEN2 p.Asn141Ile mutation, causative of autosomal-dominant Alzheimer’s disease usually by age 50 years, reported an individual with exceptional resilience in whom cognitive decline was delayed by almost 18 years.⁷⁸ This individual did not carry the RELN-COLBOS or APOE3Ch variants, and PET imaging showed less tau spreading compared with individuals with autosomal-dominant Alzheimer’s disease. Genetic analyses revealed the presence of a rare variant in the MAPT gene (MAPT p.Tyr441His), encoding for tau. Although more studies are needed to validate the genetic variants conferring resilience in this individual, these findings validate the crucial role of tau pathology in Alzheimer’s disease progression.⁷⁸

Pathways for cognitive resilience and resistance to dementia

Understanding the pathways that contribute to resilience and resistance (table) to dementia requires a multifaceted, comprehensive approach that encompasses the study of genetic, cellular, and environmental factors. Notably, the identification of protective variants in the APOE and RELN genes, which appear to influence the onset of cognitive decline in individuals with autosomal-dominant Alzheimer’s disease, suggests the existence of a crucial pathway for cognitive resilience. This pathway involves differential binding to heparan sulphate proteoglycans and ApoE receptors.^8,9 Specifically, ApoE3Ch and ApoE3-Jacksonville bind less to heparan sulphate proteoglycans, compared with ApoE4, thus supporting their protective role.^8,27 Higher ApoE4 concentrations and other diseaserelevant heparin-binding proteins have been found in plasma isolated from individuals with Alzheimer’s disease, compared with healthy control participants.⁷⁹ Conversely, the RELN-COLBOS variant seems to enhance binding to heparan sulphate proteoglycans via electrostatic interactions promoted by the presence of an additional arginine, thus leading to a gain of function.⁹ This opposite binding affinity to heparan sulphate proteoglycans of ApoE3Ch and RELN-COLBOS aligns with different binding affinities to ApoE receptors, particularly LRP-8 (or ApoEr2) and the VLDL receptor.⁸⁰

Table:

Protective genetic variants against sporadic or autosomal dominant Alzheimer’s disease

	Protective variants	Downstream pathways	Effect
ABCA7	ABCA7 rs72973581; ABCA7 Val1613Met	Lipid homoeostasis and phagocytosis, amyloid β clearance, and release of phospholipids and cholesterol31,39	Improvement of phospholipid and cholesterol transport
APOE	APOEε2	Amyloid β clearance and lipid transport23,24	Promotion of amyloid β clearance and reductions in tau pathology and amyloid plaques deposition
APOE	APOE3 V236E	Amyloid β clearance and lipid transport27,28	Reduced tau pathology
APOE	APOE3Ch	Amyloid β and tau aggregation and clearance8,11,61	Reduced tau pathology and reduced amyloid β aggregation
APP	APP Ala673Thr	APP processing44,45	Promotion of non-amyloidogenic APP processing and reduced amyloid β release
CLU	CLU rs11136000	Clusterin production and chaperone activity48,49,51	Reduced amyloid β aggregation
PICALM	PICALM rs3851179A	Amyloid β cascade47	Reduced proamyloidogenic processing and tau pathology
PLCG2	PLCG2 Pro522Arg	Antigen presentation to microglia and cascade of neuronal immune response52,53	Reduced tau pathology
RELN	RELN-COLBOS His3447Arg	Reelin signalling, brain development, and synaptic plasticity9	Reduced tau pathology and preserved synaptic plasticity
TREM2	TMEM106B; rs1990621	Microglia proliferation, survival and migration, and phagocytosis55-57	Preserved neuronal density and microglial function

The ApoE–Reelin pathway has been shown to play a crucial role in synaptic plasticity, memory formation, and the suppression of tau phosphorylation. At the cortical level, Reelin is secreted by GABAergic interneurons and binds to ApoEr2 and VLDL receptors. These receptors are mainly located in neurons involved in synaptic plasticity, such as the subgranular zone of the dentate gyrus in the hippocampus.⁸¹ This binding leads to the activation of Src kinases responsible for the phosphorylation of DAB1. DAB1 phosphorylation is a key event of Reelin signalling that leads to the activation of several downstream pathways before its ubiquitination and degradation. One of these downstream pathways involves AKT activation, which leads to inhibition of pGSK3β and results in reduced tau phosphorylation and downstream neuroprotection (figure 1). Furthermore, the observation of increased neuronal density in the entorhinal cortex of the carrier of the RELN-COLBOS variant, along with the fact that he was cognitively unimpaired, suggests that a gain of function of Reelin contributes to preserved neuronal activity and synaptic plasticity, thus resulting in enhanced memory and protection from cognitive decline.

Figure 1: The ApoE–Reelin pathway of resilience in neurons.

The intraneuronal pathway is activated upon interaction between Reelin or ApoE and heparan sulphate proteoglycans, followed by binding to the ApoE receptor. Once the Reelin–ApoE receptor complex is internalised, the rapid phosphorylation of DAB1 leads to AKT activation and subsequent inhibition of GSK3β, thus resulting in reduced tau phosphorylation. pDAB1 activity is controlled by fast ubiquitination and degradation. HSPG=heparan sulfate proteoglycans. Figure created with BioRender.com.

Tau hyperphosphorylation is crucial and correlates with cognitive impairment in individuals with Alzheimer’s disease. Given that the APOEε4 variant is a major risk factor for sporadic Alzheimer’s disease, and as the ApoE4 isoform interferes with the Reelin–ApoE pathway, a gain of function of this pathway could promote resilience against tau-induced neuropathology.⁸² ApoE4 is known to antagonise Reelin–ApoE receptor binding more strongly than other isoforms, thus supporting the hypothesis that upregulation of the Reelin pathway might be a key mechanism of resilience. Upregulation of the Reelin pathway was linked to a reduction of tau pathology in a person carrying the PSEN1 Glu280Ala variant.^8,9

The interaction between Reelin and ApoE receptors plays a pivotal role in regulating synaptic plasticity, facilitating long-term potentiation, modulating ApoE receptor functions, and promoting presynaptic vesicle release, a process also regulated by phosphorylated DAB1.⁸⁰ The protective effects of the RELN-COLBOS variant regarding increased neuronal density in the resilient individual who carried an autosomal-dominant mutation suggest the existence of a preserved neuronal network linked to the Reelin pathway and its increased binding to heparan sulphate proteoglycans, followed by a preserved interaction with APOE receptors (VLDL receptor and ApoEr2), and the consequent intraneuronal upregulation of DAB1 concentrations.⁶⁹ ApoE3Ch binds with stronger affinity to neurotoxic tau than ApoE3, thus blocking tau spreading and subsequent neurotoxicity both in culture and in two different animal models, 5xFAD and P301S transgenic mice.⁸³ These findings strongly suggest that multiple mechanisms contribute to neuroprotection in people who carry these protective variants.

On a cellular level, another possible hallmark of resilience could be the presence of abundant RORB-positive excitatory neurons, identified as susceptible to Alzheimer’s disease pathology,⁸⁴ and detected by snRNA sequencing in p-tau-protected areas, such as the frontal cortex, of the autosomal-dominant Alzheimer’s disease APOE3Ch carrier.¹⁰ The high neuronal density of layer II pyramidal neurons in the entorhinal cortex of the individual with RELN-COLBOS autosomal-dominant Alzheimer’s disease suggested the survival of neurons that are usually vulnerable by Alzheimer’s disease pathology.⁹ Reelin-positive neurons were identified as vulnerable in Alzheimer’s disease in a study using snRNA sequencing in several cortical areas of human brains of individuals ranging from 75 years to more than 90 years of age.⁷³ Thus, the discovery of genetic variants that confer resistance or resilience to Alzheimer’s disease and the identification of the specific cell populations that benefit from these protective traits present a valuable opportunity to develop treatments promoting the survival of such cells.

Neuroinflammation

Gliosis and neuroinflammation are recognised contributors to the pathogenesis of Alzheimer’s disease.⁸⁵ The protective mechanisms offered by the APOE3Ch and RELN-COLBOS variants suggest that the Reelin–ApoE pathway not only affects neurons but also modulates glial cells, such as microglia and astrocytes. On the basis of clinical observations in the APOE3Ch⁸ and RELN-COLBOS⁹ protected cases, these pathways are thought to have a crucial contribution to the preserved metabolic activity in brains resilient to Alzheimer’s disease pathology, as measured by fluorodeoxyglucose PET (figure 2).

Figure 2: Differences between individuals susceptible or resilient to Alzheimer’s disease.

(A) ApoE3 and Reelin variants contribute to maintain the homoeostatic balance between glia and neuronal transmission. In people with autosomal-dominant Alzheimer’s disease or sporadic Alzheimer’s disease, microglia, astrocytes, and synaptic transmission are severely affected. APOE4 contributes to impaired Reelin activity due to increased binding affinity to heparan sulphate proteoglycans, thus contributing to tau hyperphosphorylation. The effects of Alzheimer’s disease-inducing pathological variants are counterbalanced by the presence of protective variants such as APOE3Ch. Reelin-COLBOS contributes to preserved microglia, astrocytes, and neuronal function thanks to up-regulated reelin activity and reduced tau phosphorylation. (B) Representative fluorodeoxyglucose PET scans of the brain of an individual with autosomal-dominant Alzheimer’s disease (left) and the brain of a Reelin-COLBOS resilient individual (right) showing that the metabolic activity of the brain is severely affected in the presence of Alzheimer’s disease pathology, whereas in the presence of protective variants, metabolism is preserved. Adapted from Lopera et al.⁹ APOE3CH=APOE3 Christchurch (Arg136Ser) variant. HSPG=heparan sulfate proteoglycans. Figure created with BioRender.com.

RNA-sequencing analyses of the APOE3Ch carrier compared with non-protected PSEN1 Glu280Ala carriers have shown that ApoE3Ch elevates concentrations of the ApoE receptor LRP-1 in astrocytes in the frontal cortex, hinting at a potential resilience mechanism tied to the upregulation of LRP-1.⁵⁹ Furthermore, studies using transgenic mouse models indicate that APOE3Ch has a direct effect on microglia that results in enhanced clustering and activation, and ultimately leads to improved amyloid β clearance and reduced tau spreading.⁶⁴ Conversely, microglial ApoE4 has been shown to impair the microglial response to neurodegeneration⁶⁴ and to reduce capacity for phagocytosis and response to inflammation,⁸⁶ thus explaining its detrimental role in Alzheimer’s disease. Specifically, ApoE4 negatively affects the induction of the neurodegenerative microglia phenotype, which is associated with neuroprotection. In Alzheimer’s disease models, the deletion of microglial ApoE4 restores the neurodegenerative microglia phenotype and reduces pathology, suggesting a detrimental role of ApoE4 in microglial function.⁸⁶ In contrast to the effects of ApoE4 in lipid metabolism and microglial activation, microglia expressing the protective variant ApoE2 have been reported to do a better clearance of pathological proteins such as amyloid β, thus strengthening the role of the ApoE pathways in Alzheimer’s disease.⁸⁶

The interaction between ApoE and TREM-2 is crucial for optimal microglial function. Studies have indicated that TREM-2-mediated microglial activation enhances the beneficial effects of ApoE3, whereas deficiency of TREM-2 exacerbates the detrimental effects of ApoE4 on aberrant microglia activation.⁸⁷ Moreover, Reelin is thought to influence the differentiation of astrocytic morphology and molecular profiles according to neuronal layers via DAB1 signalling, thus ensuring their proper function.⁸⁸ Reelin might modulate the phagocytic response of microglia via ApoEr2-dependent and VLDL receptor-dependent pathways.⁸⁹ The unique neuronal density pattern seen in the RELN-COLBOS protected individual could be also interpreted as a result of the homoeostatic environment associated with a particular glial profile.

Preserved dendritic spine morphology has been reported in a cohort of cognitively unimpaired individuals with postmortem hallmarks of Alzheimer’s disease neuropathology (ie, individuals without dementia but with Alzheimer’s disease neuropathology),⁹⁰ suggesting that improved synaptic functions might mitigate tau phosphorylation and spreading. Reduced tau phosphorylation and misfolding into toxic aggregates correlate with cognitive resilience.⁹¹ A study validated that oligomeric tau is cleared from synapses in individuals without dementia but with Alzheimer’s disease neuropathology, as compared with synapses of people with sporadic Alzheimer’s disease.⁹¹ Altogether, these findings confirm the crucial role in both resilience and resistance of the pathways that contribute to low concentrations of neurotoxic tau species and preserve synaptic integrity.

Additionally, TREM-2-driven phagocytic activity is enhanced in individuals without dementia but with Alzheimer’s disease neuropathology compared with non-protected individuals with Alzheimer’s disease, which aids in the maintenance of synaptic resilience. These individuals have preserved axonal and dendritic structures near amyloid β plaques, that are known hallmarks of Alzheimer’s disease (figure 3). This preservation is associated with elevated TREM-2 expression and activity, suggesting that TREM-2 plays a protective role against synaptic loss and cognitive decline.⁹²

Figure 3: The protective role of TREM-2 in microglia.

(A) Increased TREM-2 activity leads to preserved microglia activity, whereas the crosstalk of TREM-2 with ApoE leads to preserved astrocyte function, increased phagocytosis, and preserved neuronal activity in individuals with resilience. (B) The loss of function of TREM-2 activity leads to impaired microglia activity that, along with degenerating astrocytes, contributes to neurodegeneration in people with Alzheimer’s disease. Figure created with BioRender.com.

Conclusions and future directions

Alzheimer’s disease is a heterogeneous disease that affects millions of people worldwide. Few disease-modifying treatments are available, and those that are available can cause substantial side-effects. In this Review, we have covered recent findings on protective variants against both sporadic and autosomal-dominant Alzheimer’s disease (table), with a particular focus on the variants identified in families carrying the autosomal-dominant PSEN1 Glu280Ala mutation. Notably, individuals with the APOE3Ch or the RELN-COLBOS variants had a delay in cognitive decline of almost three decades, underscoring the therapeutic potential of mapping the downstream pathways responsible for their protection.

Individuals who carry the APOE3Ch or the RELN-COLBOS variants but without dementia, and with Alzheimer’s disease neuropathology have unique protective mechanisms against either familial or sporadic forms of Alzheimer’s disease. Despite differing genetic and molecular backgrounds, these individuals share common biological traits that contribute to their protection against Alzheimer’s disease pathology, including preserved synaptic integrity, reduced tau pathology, and lower concentrations of neuro-inflammatory markers. These shared traits suggest promising therapeutic targets for developing interventions that enhance brain resilience and slow or prevent cognitive decline. Furthermore, each individual case presents a distinctive pathological profile: APOE3Ch reflects a case of resistance, given the reduced Alzheimer’s pathology and preserved cognition, whereas RELN-COLBOS is characterised by resilience. This distinction implies that there might be multiple strategies of protection against Alzheimer’s disease, which might converge on shared mechanisms. Although more research is needed to fully elucidate the molecular pathways and develop targeted treatments, these findings support the hypothesis that resilience to Alzheimer’s disease is linked to the Reelin–ApoE pathway.

Panel: Genome-wide association studies and polygenic analyses as powerful tools for the identification of genetic variants.

Genome-wide association studies (GWAS) involve the DNA analysis of a large population, which enables the identification of associations between genetic variants (ie, single-nucleotide polymorphisms) and specific diseases or phenotypes. The identification of genetic risk factors for Alzheimer’s disease and related neurodegenerative diseases is feasible through GWAS. Analyses focused on families with a particular family history of a disease are referred to as GWAX or GWAS by proxy.^13-15

Polygenic analysis involves the study of the effects of multiple genetic variants across the genome on different diseases. Polygenic risk scores are calculated as the weighted sum of risk alleles identified by a discovery GWAS.¹⁷ The resulting polygenic risk score is a parameter that allows the establishment of the genetic predisposition to a specific disease or phenotype.¹⁸

A polygenic resilience score represents the weighted sum of alleles associated with the absence of a condition in individuals with high polygenic risk scores.¹⁸ Polygenic resilience score is a novel genetic measure that quantifies an individual’s potential to resist or overcome the genetic risk for complex disorders, such as schizophrenia or Alzheimer’s disease. The identification of resilience-associated variants can inform the development of disease-modifying therapies by uncovering biological pathways involved in Alzheimer’s disease progression.¹⁶

GWAS and polygenic analyses allow identification of common genomic loci that confer either risk or resilience to sporadic Alzheimer’s disease.^13-15 The novel development of polygenic risk and polygenic resilience scores is crucial for clinical studies evaluating disease-modifying and prophylactic interventions against Alzheimer’s disease. A study from 2020 used data from four publicly available clinical cohorts (ACT, ROSMAP, ADNI, and A4) to investigate the genetic covariances between the participants’ cognitive resilience, given their amyloid burden, and 67 other complex traits, including educational attainment, mental health, and various physiological characteristics.¹⁴ The authors identified important genetic correlations between resilience phenotypes and traits, such as education, smoking behaviour, and neuropsychiatric conditions.¹⁴ Furthermore, a large GWAS involving 111 326 clinically diagnosed individuals with Alzheimer’s disease and 677 663 healthy control participants contributed to the discovery of 42 new risk loci.¹⁵ Collectively, evidence from these large-scale studies support that amyloid plaque, neurofibrillary tangle formation, cholesterol metabolism, endocytosis, phagocytosis, and innate immune system function are the pathways influencing Alzheimer’s disease risk.^15,19

Search strategy and selection criteria.

We searched the National Institute on Aging database, the ALZFORUM databases (including, AlzBiomarker, AlzRisk), and PubMed between Jan 1, 2019 and March 1, 2025 for studies describing protective genetic variants against sporadic and familial Alzheimer’s disease, using the keywords “Alzheimer’s disease” and “resilience”, “resistance”, “Christchurch”, “Reelin”, “RELN-COLBOS”, “genetic risk factors”, “pathology”, “protect”, “protective”, “variant”, “GWAS”, “TREM-2”, “pathways”, and “HSPG”. Additionally, our selection criteria aimed to help us identify current challenges in developing effective treatments for Alzheimer’s disease and elucidating pathways and genetic variants that promote resilience or resistance against the disease. We omitted studies where the full text was unavailable, preventing a thorough evaluation of their quality and relevance to the scope of this Review. Only articles published in English were consulted.