For much of modern neuroscience, the aging brain has been described through what it loses. Neurons die. Synapses become less efficient. Blood vessels deteriorate. Proteins accumulate where they should not. Memory becomes less reliable. The metaphor is usually one of gradual wear: an extraordinarily complicated machine slowly losing parts.
Two new studies suggest that this picture is incomplete.
Rather than simply deteriorating, the human brain may undergo a substantial remodeling of its immune system during later life. Cells that have lived in the brain since early development appear to decline while immune cells originating outside the brain increasingly contribute to the population that patrols neural tissue. One study identifies a particularly conspicuous transition beginning around midlife. The other provides genetic evidence that cells descended from bone-marrow stem cells actually enter the aging human brain and acquire features of microglia. (National Institutes of Health)
This is not evidence that scientists have discovered the cause of ordinary forgetfulness, dementia or brain aging. It is something more fundamental: evidence that an assumption about who inhabits the aging human brain may have been wrong.
The cells that were supposed to stay
Microglia are often called the immune cells of the brain, although that shorthand understates what they do. They move through neural tissue, sense disturbances in their environment, remove cellular debris, interact with synapses and respond to infection or injury. They are macrophages adapted to one unusually protected organ.
Their origin has made them particularly interesting.
In experimental animals, microglia arise very early in development. Precursors enter the developing brain before the mature blood-brain barrier is established. Once there, the population largely maintains itself locally. Unlike many immune cells elsewhere in the body, which are continually replenished from bone marrow, microglia have therefore been regarded as a relatively self-contained population. (Nature)
This model is well established in mice. The difficulty has always been determining whether humans follow the same rules.
You cannot label a microglial precursor in a human embryo and then inspect the same person’s brain 70 years later. Human lineage tracing therefore requires indirect methods. Two research teams have now approached that problem from different directions and arrived at strikingly compatible conclusions.
The first study, published in Science, examined postmortem hippocampal tissue from 40 neurologically healthy adults between 20 and 95 years of age. The investigators combined single-cell gene-expression measurements with epigenetic profiling and analysis of the three-dimensional organization of DNA. The hippocampus was an especially relevant target because of its central role in learning and memory and its vulnerability to age-related disease. (National Institutes of Health)
What they saw was not simply a linear progression from young cells to old cells.
Between approximately 50 and 75 years of age, the composition of the microglial population changed substantially. Cells carrying characteristics of the embryonically derived microglial population declined while another population increased. These newer cells showed molecular features resembling peripheral blood-derived immune cells and displayed stronger inflammatory signatures. (National Institutes of Health)
The important word is resembling.
From molecular profiles alone, the investigators could infer a different developmental history, but they could not directly watch those cells travel from the blood into the brain. The finding was nevertheless difficult to reconcile with the simple idea that one population of embryonic microglia quietly renews itself for an entire human lifetime.
And microglia were not changing in isolation. Cells involved in maintaining the blood-brain barrier also deteriorated with age. Across several cell types, the three-dimensional organization of the genome became less orderly, changes that were associated with altered gene regulation and cellular identity. The study therefore presents brain aging not merely as neuronal attrition but as coordinated remodeling across immune, vascular and regulatory systems. (National Institutes of Health)
Then came a second line of evidence.
Using mutations as cellular passports
In Nature, Julia Belk, Siddhartha Jaiswal and colleagues approached the problem almost like forensic genealogists.
Every time stem cells divide, they can acquire small somatic mutations. Most do nothing dramatic. But once a mutation appears in a blood-forming stem cell, descendants of that cell inherit the same molecular signature. Shared mutations can therefore act like genealogical marks connecting cells to a common ancestor.
The investigators compared immune cells from blood with cells isolated from postmortem human brains. If a microglia-like cell in the brain carried the same set of mutations as a lineage of blood cells, the probability that the match occurred independently would be extremely small. The shared mutations could instead be used to reconstruct the cells’ ancestry. (Stanford News)
Applying this strategy to 20 older individuals, the researchers found evidence of marrow-derived cells in the brains of every person examined. Single-cell analyses indicated that these newcomers could acquire molecular characteristics resembling resident microglia and, in some individuals, constitute a substantial proportion of the microglial population (Nature)
That result substantially strengthens the interpretation of the Science study.
One experiment sees an age-related cellular population with signatures suggesting a peripheral origin. The other uses naturally accumulated mutations as lineage markers and traces brain immune cells back toward blood-forming stem-cell populations. Different methods, different samples, similar biological direction (National Institutes of Health)
The Stanford group also reported that this phenomenon does not appear to occur in the same way in mice or nonhuman primates. If confirmed, that observation matters enormously because much of what neuroscience knows about microglial aging comes from experimental animals. A biological process that becomes prominent in aging humans but is absent or much weaker in standard laboratory species could be systematically underestimated by conventional models. (Stanford News)
Aging as replacement rather than simple decay
The conceptual change is subtle but important.
Consider two models of an aging city.
In the first, the same inhabitants and institutions remain in place while infrastructure gradually deteriorates. In the second, deterioration is accompanied by migration: an old workforce declines, newcomers arrive and the character of the city changes because its population has changed.
Brain aging may contain elements of both.
Neurons, vasculature, proteins and gene regulation still undergo well-documented age-related alterations. These findings do not overturn that biology. But they suggest that part of the aging environment may also arise because the cellular composition of the brain itself is changing.
That could have consequences for neuroinflammation.
The microglia-like cells identified in the Science study showed stronger inflammatory characteristics than the population they appeared to replace. Chronic low-grade inflammation is common in aging and is strongly implicated in several neurodegenerative disorders. A changing immune-cell population could therefore contribute to the inflammatory environment of an older brain. (National Institutes of Health)
But this is precisely where caution becomes essential.
An inflammatory molecular signature does not establish that these cells cause cognitive decline. Nor does their appearance around the same period when dementia risk begins rising rapidly establish causality. They could contribute to degeneration. They could arrive in response to damage. They could compensate for failing resident microglia. Different subpopulations could even do different things.
The Nature study provides a particularly intriguing reason not to assume that blood-derived cells are harmful. Its analysis found that several forms of clonal hematopoiesis — age-related expansions of particular blood stem-cell clones — were associated with lower risk of Alzheimer’s disease. The biological explanation remains unresolved, but the result raises the possibility that at least some marrow-derived immune populations entering the brain may be protective rather than destructive (Nature)
The immune remodeling of the aging brain may therefore not be simply another form of damage. It may represent adaptation, maladaptation or some mixture of both.
The blood-brain barrier looks less like a wall
The findings also complicate one of neuroscience’s most persistent metaphors: the brain as an immunologically secluded organ protected behind the blood-brain barrier.
The barrier remains biologically crucial. It regulates molecular and cellular traffic between the circulation and neural tissue. But barrier should never have been interpreted as absolute wall.
If immune cells derived from blood-forming stem cells increasingly populate the aging brain, the biological history of the bone marrow and circulation becomes potentially relevant to the biological history of the brain. (Stanford News)
That creates an unexpected bridge between hematology and neuroscience.
Blood stem cells accumulate mutations with age. Their clonal composition changes. They are influenced by systemic inflammation, cardiovascular disease, infection and other exposures. If descendants of these cells eventually enter the brain and become microglia-like cells, events occurring far from the skull could potentially alter the immune ecology of neural tissue decades later.
That possibility remains a hypothesis, not a demonstrated causal chain. But it opens an unusually fertile research direction: brain aging may partly reflect the cumulative history of the body’s hematopoietic system.
What this could eventually mean
There is an obvious therapeutic temptation.
The blood-brain barrier has long frustrated attempts to deliver treatments to the central nervous system. If certain peripheral immune cells naturally enter the aging brain, researchers may eventually ask whether those cells could be engineered before they arrive.
Belk and colleagues have raised the possibility of modifying peripheral immune cells so that, once inside the brain, they could perform useful functions — potentially enhancing clearance of pathological proteins such as amyloid or tau. (Stanford News)
That is an intriguing idea. It is also far from a treatment.
The current studies involve observational analyses of human tissue. They do not demonstrate that manipulating this migration prevents dementia, preserves memory or reverses neurodegeneration. They do not tell us whether more infiltration is better or worse. They do not establish an optimal cellular phenotype. Even the apparent age window of 50 to 75 years should not be interpreted as a biological deadline: it emerged from cross-sectional postmortem data, not repeated measurements of the same individuals across decades. (National Institutes of Health)
Perhaps most importantly, the Science study focused on the hippocampus. It does not establish that every region of the human brain undergoes the same transition at the same age.
These limitations make the findings more interesting, not less. Good biological discoveries often begin by revealing that a supposedly settled assumption contains an exception.
For decades, researchers asked how the brain’s resident immune cells change as they age.
The emerging question is different.
Are they always the same cells?
The answer now appears to be: not necessarily.
And if the cellular guardians of the brain are progressively replaced by descendants of the immune system circulating through the rest of the body, then aging of the brain cannot be understood entirely by studying the brain in isolation.
After 50, the story may not simply be that neurons are growing older.
The population around them may be changing too.
Primary studies:
Nathan R. Zemke et al. Epigenetic and 3D genome reprogramming during the aging of the human hippocampus. Science. 2026. DOI: 10.1126/science.adt8307. (National Institutes of Health)
Julia A. Belk et al. Somatic mutations reveal the ontogeny of microglia in human aging. Nature. 2026. DOI: 10.1038/s41586-026-10939-0. (Nature)
