Summary: Researchers have identified 216 gene variants that appear to influence the development of late-onset Alzheimer’s disease.
Source: Baylor College of Medicine
Late-onset Alzheimer’s disease (LOAD) is the most common form of dementia among older adults, affecting tens of millions worldwide. Despite extensive research, the genetic contributors that increase or reduce risk for LOAD remain incompletely understood. As populations age globally, there is an urgent need for more accurate prognostic biomarkers and new therapeutic targets.
Besides advanced age, variants in the apolipoprotein E (APOE) gene are the strongest known genetic predictors of Alzheimer’s risk. The APOE ε4 allele raises the likelihood of developing the disease, whereas the ε2 allele generally confers protection. Yet many individuals do not follow this pattern: some APOE ε4 carriers remain healthy into old age, and some APOE ε2 carriers still develop Alzheimer’s. Understanding these “rule-breakers” could reveal genetic modifiers that change disease course and point to new biomarkers or drug targets.
Using an innovative approach that integrates evolutionary information, machine learning and high-throughput experimental validation, researchers at Baylor College of Medicine and the Jan and Dan Duncan Neurological Research Institute at Texas Children’s Hospital analyzed the genomes of these paradoxical individuals. Their work uncovered 216 genes with variant patterns that differ significantly between APOE ε2 and ε4 paradoxical groups. Many of these genes were not previously linked to Alzheimer’s disease.
Study leadership and publication
The study was led by Drs. Olivier Lichtarge and Juan Botas, professors of molecular and human genetics at Baylor College of Medicine and investigators at the Duncan NRI. The full report appears in the journal Alzheimer’s & Dementia.
“We set out to test whether people who show paradoxical outcomes carry other genetic variants that blunt or reverse the expected effects of their APOE genotype,” Lichtarge said. Graduate student Young Won Kim developed computational methods to compare genetic variation across individuals and prioritize candidate modifier genes. The team then used a unique, high-throughput fruit fly screening platform at the Duncan NRI to test biological effects experimentally.
A computational strategy reveals 216 candidate genes linked to paradoxical APOE outcomes
Because LOAD presents with wide clinical variability, complex inheritance, and frequent co-occurrence with vascular and other age-related conditions, typical genetic studies can miss important modifiers. The investigators focused on the Alzheimer’s Disease Sequencing Project (ADSP), the largest available whole-exome dataset for this condition, and restricted their analysis to coding regions where variants are most likely to alter protein function.
To estimate the functional impact of variants, the team incorporated evolutionary information—measuring past conservation and divergence at each gene position—and carried out regression analyses to compare mutational burdens between paradoxical APOE ε2 and ε4 groups. This approach identified 216 genes with significant differences in variant load. Many of these genes participate in pathways already implicated in Alzheimer’s biology, including synaptic function, dendritic spine pruning, and neuroinflammation.
Using machine learning models trained on the variant profiles of these 216 genes, the researchers could predict which APOE ε4 carriers were likely to remain healthy and which APOE ε2 carriers were at greater risk of developing Alzheimer’s, suggesting potential utility for improved risk stratification.
High-throughput fruit fly screen validates biological effects
To test whether candidate genes actually modify Alzheimer’s-related neuronal dysfunction, the team turned to an experimental pipeline built around Drosophila models engineered to exhibit Alzheimer’s pathology. Leveraging custom robotic systems developed at the Duncan NRI, the group performed a high-throughput behavioral screen that quantitatively measures nervous system function—primarily locomotor behavior—as a readout of neuronal health.
These robotic assays allowed the researchers to assess how altering each candidate gene affected neuronal performance, and to distinguish whether variants caused loss-of-function or gain-of-function effects. That distinction is crucial for designing downstream therapeutic strategies, because interventions may need to either enhance or inhibit the activity of specific targets.
“By concentrating on paradoxical patient cohorts, we can uncover genetic modifiers that allow some people to defy expected disease trajectories,” said Dr. Ismael Al-Ramahi, co-lead author. “This study provides proof of concept that blending evolutionary analysis, machine learning and high-throughput Drosophila genetics can reveal targets that refine risk prediction and point to potential therapeutic directions.”
The authors note that leveraging evolutionary information partly compensated for the relatively small sample size used here (about 480 individuals), compared with much larger genome-wide association studies. That suggests this study design could be an efficient way to study other conditions where “rule-breaking” patients exist.
Other contributors include Amanda Koire, Stephen J. Wilson, Daniel M. Konecki, Samantha Mota and Shirin Soleimani, affiliated with Baylor College of Medicine and the Jan and Dan Duncan Neurological Research Institute at Texas Children’s Hospital.
Funding: This research was supported by the National Institutes of Health and the Oskar Fischer Foundation.
About this Alzheimer’s disease research news
Source: Baylor College of Medicine
Contact: Jenn Jacome – Baylor College of Medicine
Image: The image is in the public domain
Original Research: The study will appear in Alzheimer’s & Dementia