The ravages of age touch us all — some more lightly than others. Geneticist Henne Holstege wanted to study people with a particular superpower: those resistant to neurodegeneration, despite their age or their genes putting them at high risk.
At first, she struggled to get funders interested. “For five years, all my grants were rejected,” she recalls. “Each time was a cup of poison.”
But she persevered, convinced that these super-agers could help scientists to understand why a privileged few people manage to avoid conditions such as neurodegeneration, which ail so many in older age.
She finally launched her project in 2013, and ever since, Holstege and others in her field have uncovered a rash of gene variants that seem to protect against dementias.
Researchers are increasingly convinced that understanding resilience to dementia is just as interesting as the more familiar genetics of high risk.
The brain, it seems, has its own defence mechanisms against the toxic proteins that cause Alzheimer’s — mechanisms that scientists now hope to exploit.
They think that protective gene variants will help to illuminate how the disease emerges and worsens — and potentially how to treat it. Of particular interest is a group of variants that affect the workings of the ApoE protein (apolipoprotein E), which carries fat molecules essential for cellular function. Some forms of this protein are long-known contributors to the risk of Alzheimer’s. But some recently discovered and rare variants seem to confer resistance.
“Nature has already done the experiment for us: certain gene variants protect us from Alzheimer’s and they do so safely,” says neuroscientist Guojun Bu from the Hong Kong University of Science and Technology. “Learning how they do this could guide us to new therapies.”
Impervious to plaques
The relentless course of Alzheimer’s disease follows a decades-long progression that is kick-started when amyloid-β proteins in the brain clump into plaques. It concludes with loss of synapses and death of brain cells.
Except when it doesn’t. Just occasionally, people whose brains are full of plaques don’t follow the usual route. They stay free of dementia for most or all of their long lives. A few people don’t even build plaques, even though some might carry gene variants conferring a high Alzheimer’s risk.
In the normal course of disease, the amyloid-β proteins in plaques activate immune cells in the brain known as microglia, which try to clear them. But if active for too long, the microglia become inflammatory and start to damage neurons. This, along with other cellular stresses, promotes the build-up and spread of a different, deadlier type of protein, called tau, which forms long, thin tangles. These tangles correlate with the development and severity of dementia symptoms.
Although lifestyle and environmental factors can affect the likelihood of developing the disease, the most powerful influence is genetics.
The exact mechanistic details of how plaques lead to symptoms are still unclear. Gene variants that either accelerate or slow disease could affect many pathways. One such pathway might relate directly to clearance of toxic amyloids or spreading of tau, but could also be acting indirectly, by affecting energy systems in brain cells, for example. “That’s why it is important to identify these variants and discover their exact mechanisms,” says Holstege, who is at Amsterdam University Medical Centre.
Searching the genome
The search for gene mutations that contribute to the risk of Alzheimer’s goes back decades. A small number of Alzheimer’s cases are strongly hereditary, involving mutations in genes that result in a higher proportion of stickier, toxic forms of amyloid-β proteins. Just one copy of any of these mutated genes almost inevitably leads to disease development early in adult life, most often when someone is in their 40s or 50s.
However, familial early-onset Alzheimer’s accounts for fewer than 1% of all cases; most cases affect people over 65. The APOE gene is the most important genetic determinant of the late-onset form of disease. The protein it encodes, ApoE, is central to lipid transport between cells and around the body. If it does its job properly, it can help to prevent amyloid build-up and clear it from the brain; it can also influence tau pathology, partly through effects on microglia.
Three main alleles of the gene exist, which generate a spectrum of risk. The alleles differ from each other in only one or two nucleotide positions. APOE3 is the most common variant. APOE4 dials the risk up and is the greatest genetic risk factor known. “The particular nucleotide substitution makes the protein it encodes more likely to aggregate and less likely to do its day job of transporting lipids,” says Bu. APOE2 dials the risk down: two copies can delay onset of Alzheimer’s by a decade. Only about 1% of the population has two copies of APOE2.
But however notable the effects of APOE alleles, the vast majority of late-onset Alzheimer cases result from the combined effects of many, often rare, gene variants — some increasing and others reducing disease risk — as well as short stretches of DNA that influence how much each gene is expressed.
Scientists have so far identified only part of the genetic contribution to Alzheimer’s risk. They think that the missing genetics — including numerous but rare variants — might hold some of the most biologically informative clues about resilience.
The usual way to search for variants is in genome-wide association studies (GWAS), in which scientists analyse millions of spots on the genome for differences between groups — for example, those with disease or at high risk of disease, compared with those who are disease-free or cognitively resilient. These spots are called single nucleotide polymorphisms (SNPs). A SNP can act as a flag for a stretch of DNA that looks different in high-risk individuals than in those who are cognitively resilient.
To search for extremely rare resilience genes, you need sequences from huge numbers of people and sensitive identification techniques. Both have improved in the past decade or so. Hundreds of thousands of genomes from people with and without Alzheimer’s are available from sources such as UK Biobank. “The more people you look at, the more likely you are to spot very rare occurrences,” says Michael Greicius, a geneticist at Stanford University in California.
New cohorts are being put together all the time, and those that include people of Asian, African and other ancestry show significant differences in Alzheimer’s risk genetics compared with predominantly European-ancestry cohorts1.
Greicius’s team has found a rare protective variant known as W276C in a population of people with Native American ancestry, which has not been seen in the other cohorts2. “There’s an incredible amount of power in ancestrally diverse data sets that we miss completely when we myopically focus on European ancestry,” he says.
Not-so-sticky proteins
Once candidate resilience variants have been identified in human genetic studies, scientists begin the lengthy task of working out whether — and how — they might affect disease biology. They mutate these genes in mice or cell cultures and also look for cells or biological pathways that have been affected in donated brain tissue.
A dozen or so rare gene variants have now been verified in this way as likely contributors to Alzheimer’s resilience, although the strength of evidence varies greatly between them. APOE itself has produced some of the clearest examples. Three extremely rare variants of APOE3 have been found in people who seem to escape Alzheimer’s, and one variant of APOE4 seems to neutralize the risk usually carried by that form. All of them are gold dust for scientists. “These rare variants hold the key to understanding how the common risk variants do what they do,” says biochemist Rosie Jackson at the University of Dundee, UK.
The changes to the protein might be tiny, but the consequences are huge. They all affect how the ApoE protein binds to molecules.
ApoE is shaped a bit like a hairpin (see ‘Protective protein’). One arm has surfaces that attract and carry lipids. Once a lipid is bound, the protein opens up and reshapes itself around it. The other arm binds to receptors on cells, allowing those cells to ingest ApoE and use its lipid cargo. These receptors can also take up tau and contribute to its spread between cells.

Sources: T. Kanekiyo et al. Neuron 81, 740–754 (2014); H.-J. Moon et al. Front. Mol. Neurosci. 17, 1399965 (2024).
There are three rare variants found in the APOE3 allele. One is known as the Christchurch variant (after the city in New Zealand in which it was discovered) and it affects the receptor-binding region of the ApoE protein. It is thought to bind ApoE less strongly to receptors, making tau less likely to spread.
The other two rare APOE3 variants influence the lipid-binding section, probably making the protein less sticky and less likely to clump and form plaques.
One of these is W276C, found in the Native American-ancestry population. The other is known as the Jacksonville variant and was first reported3 in 2014 by a team that Bu was part of at the Mayo Clinic in Jacksonville, Florida. The group used information from a public genetic database to identify rare changes to Alzheimer’s genes. Then they looked at the genomes of thousands of people with and without Alzheimer’s to see how often it was present in each.
In subsequent work, Bu’s team introduced this variant into mice that were engineered to develop Alzheimer’s-like amyloid pathology4. The animals developed fewer plaques and less neuronal damage around the plaques. They had increased levels of several lipid classes that are important for synaptic function. The team also inserted the Jacksonville substitution into the APOE4 gene and found in biochemical experiments that it significantly reduced the ApoE4 protein’s tendency to aggregate.
The human genetic evidence for Jacksonville was strengthened in a 2022 study5 in which a team analysed genetic data from more than half a million people and found hundreds of carriers of the variant. That study also identified a protective variant of APOE4 called R251G, now the subject of several ongoing studies.
The Christchurch variant has the least human genetic support; just one person with two copies of the variant has been identified6. Despite a strong genetic predisposition to early-onset Alzheimer’s, she remained cognitively healthy for three decades longer than expected, even though her brain contained large amounts of amyloid plaques. (Having one copy of the variant doesn’t seem to offer much protection.)
Nonetheless, laboratory experiments have provided strong evidence that the Christchurch variant can change ApoE biology in ways that could plausibly be protective — reducing plaque burden or changing how cells handle tau.
David Holtzman’s team at Washington University in St Louis is one of the groups working on this. In a study7 using cell lines, his team found that the Christchurch variant substantially reduced the amount of tau that developed around plaques and that this depended in part on microglia, which took up more tau and broke down tau aggregates.
Neuroscientist Yadong Huang at the University of California, San Francisco, decided to insert the Christchurch mutation into mice carrying the human APOE4 gene8. “We wanted to see if that little change would stop its detrimental effect.” Two copies strongly reduced the tau pathology, neurodegeneration and neuroinflammation that APOE4 would normally drive. One copy partly protected against neurodegeneration and neuroinflammation, but seemed not to be powerful enough to reduce tau pathology itself.
All of this work has revealed much about the complexity of ApoE’s biology, and points to approaches to drug development beyond the current focus on reducing amyloid and tau. “So far, there has been almost no effort in the pharmaceutical industry to develop treatments based on ApoE — unbelievable given how strong its impact is,” says Holtzman. The rare variants are helping to identify druggable targets, he says.
Beyond ApoE
ApoE isn’t the only game in town: other genes carry candidates for rare protective variants, although research on them is less advanced. The first to be found was a mutation discovered back in 2012 in an Icelandic cohort9. It is in the gene that encodes the amyloid precursor protein, which is chopped up by an enzyme into the sticky amyloids that form plaques. The variant makes it harder for the enzyme to do this10.
Another, reported in 2024, pointed to a previously underexplored biological pathway11. The researchers studied cohorts of people who remained cognitively healthy despite carrying two copies of APOE4. They identified rare variants in the gene for fibronectin 1, a protein that provides structural support for cells and blood vessels. “It was surprising at first to see this gene pop up,” says neurologist Richard Mayeux at Columbia University, New York City, one of the study’s leaders. It took them two years to work out how the variant was protective. They discovered that APOE4 was causing too much fibronectin to accumulate around blood vessels in the brain, contributing to the leakiness of the blood–brain barrier seen in Alzheimer’s12. The protective variants made less fibronectin, therefore limiting the build-up.
In the complex world of human biology, multiple variants probably interact to determine an individual’s brain health. Some labs have embarked on ambitious programmes to build detailed maps of how variants alter molecular processes in specific brain-cell types and brain areas.
That’s going to be very revealing, says Mayeux. “But the main thing is to try and figure out whether there is something we can put into a human which will slow the process of Alzheimer’s down,” he says. “That’s everybody’s goal.”
The rare variants are flagging up points in the development of Alzheimer’s that could serve as druggable targets, says Laura Nisenbaum at the New York City-based Alzheimer’s Drug Discovery Foundation, which funds translational research.
A lot more research is needed to understand how to develop them, she says, noting that it will take many more years to move from a validated target to a marketed drug. But the key selling point is that scientists already know they are protective. “A rare variant is basically nature showing us a drug target that already works in a human being.”
References
- Lee, S. et al. Alzheimers Dement. 20, 3397–3405 (2024). doi:10.1002/alz.13796
- Le Guen, Y. et al. Preprint at medRxiv (2026). doi:10.64898/2026.05.08.26352778
- Medway, C. W. et al. Mol. Neurodegener. 9, 11 (2014). doi:10.1186/1750-1326-9-11
- Liu, C.-C. et al. Sci. Transl. Med. 13, eabc9375 (2021). doi:10.1126/scitranslmed.abc9375
- Le Guen, Y. et al. JAMA Neurol. 79, 652–663 (2022). doi:10.1001/jamaneurol.2022.1166
- Arboleda-Velasquez, J. F. et al. Nature Med. 25, 1680–1683 (2019). doi:10.1038/s41591-019-0611-3
- Chen, Y. et al. Cell 187, 428–445 (2024). doi:10.1016/j.cell.2023.11.029
- Nelson, M. R. et al. Nature Neurosci. 26, 2104–2121 (2023). doi:10.1038/s41593-023-01480-8
- Jonsson, T. et al. Nature 488, 96–99 (2012). doi:10.1038/nature11283
- Shimohama, S. et al. J. Neurosci. 44, e0223242024 (2024). doi:10.1523/JNEUROSCI.0223-24.2024
- Bhattarai, P. et al. Acta Neuropathol. 147, 70 (2024). doi:10.1007/s00401-024-02721-1
- Bhattarai, P. et al. Nature Aging 6, 1901–1927 (2026). doi:10.1038/s43587-026-01204-0
Shared from Nature, for the original and more great content visit: doi: https://doi.org/10.1038/d41586-026-03625-8
