Tiroyaone Brombacher sat in her lab at the University of Cape Town watching a video of an albino mouse swimming around a meter-wide tub filled with water. The animal, which lacked an immune protein called interleukin 13 (IL-13), was searching for a place to rest but couldn’t find the clear plexiglass stand that sat at one end of the pool, just beneath the water’s surface. Instead, it swam and swam, crisscrossing the tub several times before finally finding the platform on which to stand. Over and over, in repeated trials, the mouse failed to learn where the platform was located. Meanwhile, wildtype mice learned fairly quickly and repeatedly swam right to the platform. “When you took out IL-13, [the mice] just could not learn,” says Brombacher, who studies the intersection of psychology, neuroscience, and immunology.

Curious as to what was going on, Brombacher decided...

As far back as 2004, studies in rodents suggested that neurons and their support cells release signals that allow the immune system to passively monitor the brain for pathogens, toxins, and debris that might form during learning and memory-making, and that, in response, molecules of the immune system could communicate with neurons to influence learning, memory, and social behavior. Together with research on the brain’s resident immune cells, called microglia, the work overturned a dogma, held since the 1940s, that the brain was “immune privileged,” cut off from the immune system entirely. 

INTO THE BRAIN: Astrocytes (green) have long, thin extensions called feet that line the blood vessels in the brain. Those feet contain water channels (purple) that allow cerebrospinal fluid, including the molecules within it, to move into the astrocytes and out into the brain, where it can interact directly with neurons.

Brombacher and others are now starting to identify how communication between the nervous system and the immune system happens. In 2012, molecular imaging revealed that fluorescently labeled proteins could flow through a layer of projections, or “feet,” of neuronal support cells called astrocytes. Astrocytes are star-shaped cells that sit at the border of neural and meningeal tissues and along the blood vessels of the brain; their foot layer is the barrier that separates cerebrospinal fluid (CSF), the watery liquid that envelops the brain and spinal cord, from the neurons of the central nervous system. If those fluorescently labeled molecules could cross the astrocyte layer and move into and out of the brain, so could CSF-based immune-system proteins, which are smaller, scientists figured.

Experiments have also shown that cytokines in the blood can cross the blood-brain barrier (BBB—which, in addition to the wall of astrocyte feet, includes a tight layer of endothelial cells surrounding the brain’s vasculature—and may influence neurons. A third mode of communication, Brombacher notes, is through immune cytokines’ interactions with astrocytes themselves: it seems that the signaling molecules don’t have to penetrate neural tissue at all to influence the brain. Her work shows, for example, how cytokines such as IL-13 spur astrocytes to release brain-derived neurotrophic factor (BDNF) and other proteins that bolster neural development and influence learning and memory.

This line of work has led to rapid developments in neuroimmunology, a growing field of research that focuses on understanding the ways in which the nervous system draws on immune cells during normal function, and how that interaction plays a role in learning, memory, and social behavior, as well as neurological disease. Some researchers even propose that the immune system might be key to treating some forms of impaired cognition.

Cytokines that influence cognition

One of the first teams to make the connection between the immune system and brain function included Jonathan Kipnis, now at the Washington University School of Medicine in St. Louis. In 2004, Kipnis and colleagues showed that mice without adaptive immune cells such as T cells had trouble remembering the location of a submerged platform while they were swimming. A few years later, the group focused in on a T cell cytokine called interleukin-4 (IL-4), which helped mice with functional immune systems form long-term memories about the platform’s location. IL-4 is secreted by T cells in the body that can migrate to the meninges, and was somehow affecting the brain.

Following up on that work, Kipnis’s then-postdoc Anthony Filiano, now an assistant professor of neurosurgery at Duke University, found that mice lacking T cells didn’t socialize with others the way normal mice did. If the immune-deficient mice got an infusion of immune cells at around four weeks of age, they became much more social, mimicking the behaviors of normal mice just a few weeks after their immune supplementation. An analysis of gene expression data collected from both sets of mice revealed that interferon gamma, a cytokine essential for the body’s defense against viral and bacterial pathogens, was associated with sociality. 

To see if interferon gamma had a direct effect on the brain, Filiano and his collaborators knocked out the gene for the cytokine receptor in neurons in the mouse prefrontal cortex, a region important for social behavior. This caused mice to spend less time interacting with other mice, a sign that they were feeling less social; the result offered evidence to suggest that interferon gamma from T cells in the meninges was acting directly on the cortical neurons.

Inspired by Kipnis and Filiano’s work, Brombacher and colleagues decided to set up a similar experiment. The team first tested IL-4 knockout mice against wildtype mice in a water maze and successfully replicated Kipnis’s original results—the immunodeficient mice were learning impaired. Then, Brombacher tried the experiment with mice lacking IL-13, which is closely related to IL-4, and got the more dramatic results: “learning was abrogated,” she says. Both cytokines clearly affected learning, but IL-13 appeared to play a more significant role than IL-4, perhaps because of some underlying biochemistry: IL-13 and IL-4 share a receptor on the surface of cells called IL-4 receptor alpha, but IL-13 can also transmit its signal using another receptor. Brombacher is now setting up experiments to remove the cell receptors and see what happens to the mice’s performance in the water maze.

Turning to the immune system might be key to treating some forms of impaired cognition.

Evidence is also mounting for interleukin 17’s involvement in learning and sociality. In 2016, Gloria Choi of MIT’s McGovern Institute for Brain Research and colleagues linked the cytokine to signs in mouse pups similar to symptoms of autism spectrum disorder (ASD) in humans. Specifically, animals that developed infections while pregnant gave birth to babies that exhibited ASD-like behavioral traits. Interleukin 17 (IL-17) was among the immune signals secreted to help combat the pathogen, the researchers found, and baby mice born to mouse moms with infections had a higher abundance of IL-17 receptors on their brain cells than mice born to uninfected moms. Blocking those IL-17 receptors with drugs during gestation protected pups against the effects of higher maternal IL-17; the pups were born without the signature behavioral issues associated with ASD. Stimulating the release of IL-17 or administering the cytokine directly also appeared to attenuate ASD-like symptoms in young and old adult mice that had been exposed to the high IL-17 levels in utero, suggesting that exposure to elevated maternal IL-17 during development also paradoxically “primes” the immune system for rescue by the cytokine in maturity.

Last year, neuroimmunologist Julie Ribot of the University of Lisbon and her colleagues added to the IL-17 story when they discovered that mice lacking a certain type of T cell or the cytokine had trouble making short-term memories when exploring a Y-shaped maze. This is in contrast to the effects on long-term memory formation that researchers have uncovered for IL-4 and IL-13 in the water maze. The different effects of the interleukins, Ribot says, could have something to do with the fact that the gamma delta T cells that produce IL-17 reside in the meninges, where they could act within seconds during short-term memory formation. T cells that produce IL-4 and IL-13, on the other hand, have to be recruited to the meninges from elsewhere in the body, which takes time, suggesting they support the creation of memories that take longer to form, she notes.

The role of gamma delta T cells and IL-17 in cognition doesn’t end with links to memory and autism, though. The cells and their cytokine may play a role in anxiety, according to Kipnis’s latest experiments. His team recently showed that the release of IL-17 from gamma delta T cells correlates with anxiety-like behavior in mice, and that deleting the IL-17 receptor from glutamatergic neurons in cortical regions involved in threat perception and response reduced anxiety-like behaviors. The major takeaway from each of these IL-17 papers is the same, Kipnis says. “We’re showing that you have a population of immune cells sitting outside the brain that impact neurons inside it.”

In parallel with these studies demonstrating the capacity of cytokines to affect learning and memory, anxiety, and social behavior, researchers are beginning to pull back the curtain on the communication channels that T cells use to talk with neurons. Although it is still unclear exactly how the two systems physically interact, several possibilities have been identified, including direct messages from T cells sent via cytokines interacting with neurons and indirect signals generated through the interaction of cytokines with astrocytes.

Immune-Neural Crosstalk

Several routes exist for immune cells and neurons to communicate, though T cells rarely come in direct contact with neural tissue. This communication can happen as cerebrospinal fluid (CSF) flows from the space surrounding blood vessels deep in the brain into neural tissue and back out again. As an animal learns new information, changing neural circuits can release signals to which the immune system responds. The immune system in the meninges, the spongy membranes that separate neural tissue from the skull, also monitors CSF coming from the brain for signs of infection or injury.


The meninges’ innermost layer, the pia mater, lines the perimeter of the brain, separating neural tissue from the surrounding fluid and tissue. But gaps in the thin, fibrous tissue allow blood vessels to extend deep into the brain. 

Along blood vessels in the brain, a tightly packed layer of endothelial cells, along with projections, or “feet,” from astrocytes collectively make up the blood-brain barrier, which prevents blood from entering the organ. But CSF that sits in the space between the pia mater and upper layers of the meninges flows down around the endothelium-lined blood vessels.

As arteries pulse with each beat from the heart, CSF pushes into the astrocyte feet through AQP4 water channels. This CSF can carry signals from the immune system such as cytokines IL-17, IL-4, and interferon gamma that may also talk directly with neurons.

Cytokines can also trigger astrocytes to release molecules such as brain-derived neurotrophic factor (BDNF), influencing learning, memory, and sociality.

Once in the brain, the CSF mixes with extracellular fluid from neuronal tissue, sweeping up cellular waste excreted along with any toxins, pathogen-derived antigens, and debris formed as part of normal neural rewiring. This fluid is then pushed out of the brain through astrocyte feet into the perivascular space, where it can interact with gamma delta T cells. Those T cells may then respond by releasing cytokines such as IL-17 that can move right back into the brain, although this has yet to be shown.

The CSF is then channeled to the lymphatic vessels in the meninges and flushed to lymph nodes in the neck, where more T cells are waiting to scan the fluid and respond.

See full infographic: WEB | PDF

Communication lines between immune cells and neurons

Some neuroscientists remain adamant that, with the exception of some drugs, most molecules do not get through the barriers that separate the brain from the rest of the body unless there’s a rupture to the boundary layers intended to cordon off the central nervous system. But research from several groups now challenges this idea. A key study in disproving the long-held assumption that the brain is immune privileged came from the lab of neuroscientist Maiken Nedergaard of the University of Rochester Medical Center. In 2012, she and her colleagues watched fluorescent and radiolabeled tracers flow from the CSF into the brains of anesthetized mice. 

See “Getting Drugs Past the Blood-Brain Barrier

Specifically, Nedergaard’s team recorded the movement of the tracers into and out of the animals’ cerebral cortex, the brain’s outer layer of folded gray matter, which is essential for consciousness, attention, and making memories. The researchers learned that CSF carrying cytokines and other signaling molecules flows from the meninges into the space surrounding the brain’s vasculature. As the arteries pulse with each beat of an animal’s heart, the blood vessels expand, and the CSF is pushed through water channels in the astrocyte feet and then into the brain. The reverse flow also takes place: CSF that has entered the brain and mixed with the extracellular, or interstitial, fluid—and that now carries waste proteins ready for clearance—is pressed back through astrocytes into the space surrounding the blood vessels. “Maiken showed this very, very elegantly,” Kipnis says. It completely overthrew the dogma that the brain is immune privileged, he says. 

Earlier this year, Andrew Yang of Stanford University and colleagues extended this finding to specifically show that cytokines released from T cells in the blood can also reach the brains of mice. The researchers extracted blood from the animals, separated out plasma proteins, labeled them with a fluorescent tag, then injected them back into the bloodstreams of the mice they came from. In healthy young adult mice, lots of the fluorescently tagged proteins crossed the BBB to enter the interstitial fluid in the brain. “This finding suggests that a wide variety of neural functions . . . could be modulated by systemic protein signals,” Roeben Munji and Richard Daneman of the University of California, San Diego, wrote in a commentary accompanying Yang’s study.

Cytokines in the meninges or possibly even in the blood might not have to enter the brain at all to affect the central nervous system, according to Brombacher’s studies. IL-13 and other cell signaling molecules in the CSF or blood could interact with astrocytes at the BBB or at the perimeter of the brain. In cultured astrocytes, treatment with IL-13 spurred the production of BDNF and triggered the production of glial fibrillary acidic protein (GFAP), an indication that neural connections are undergoing rewiring. Ribot’s study on IL-17 also showed that the cytokine could spur mice’s astrocytes to release BDNF into the brain. Both BDNF and GFAP, which boost synaptic rewiring, are associated with learning and memory.

Because we know that cerebrospinal fluid does go into the brain, putting therapies into that fluid will probably be a very, very efficient route for treating patients.

 —Jonathan Kipnis, Washington Univer­sity School of Medicine 

Communication between the immune and nervous systems can also happen in the reverse direction, with signals from the brain reaching T cells of the spongy, membranous meninges and of the rest of the body. Until a few years ago, researchers agreed that the brain lacked a drainage, or lymphatic, system to clear away its waste and to transport immune cells. But to Kipnis, it just didn’t make sense that one of the most important organs in the body would not have that kind of plumbing. So he and his colleagues went looking for it, and in 2015, they found it—mice’s brains did, in fact, have lymphatic vessels that shipped waste and T cells from the meninges to deep cervical lymph nodes in the animals’ necks. “These structures are bona fide vessels—they express all the same markers as lymphatic vessels in every other tissue,” he told The Scientist at the time. 

In mice, T cells residing in the meninges scan the CSF for the cellular waste generated as neuronal circuits undergo changes, whether in response to learning and memory formation or in the case of dysfunction. Then, T cells in the lymph nodes get a chance to check the CSF for potential threats such as pathogens, he says. Tracking that lymphatic system in people’s brains is much harder, but Kipnis says there’s some evidence that what scientists are finding in rodents does translate to human biology. In 2017, he and collaborators at the National Institute of Neurological Disorders and Stroke used MRI to noninvasively confirm the existence of meningeal lymphatic vessels in humans and nonhuman primates.

Understanding immune cell–neuron crosstalk—both the way T cells respond to what’s in CSF coming from the central nervous system and how they send signals into the brain—could be important for understanding neurological disorders, such as Alzheimer’s disease, autism, schizophrenia, and even the cognitive decline associated with aging. “With many of these neurological disorders, there’s been reports that there’s some kind of dysregulation of the immune system,” Filiano says. Identifying faulty signals from neurons in the fluid leaving the brain could lead to diagnostic tools for neurological disorders, he notes. And given that CSF can carry cytokines and other proteins to neurons, Kipnis says he suspects that “putting [immune-based] therapies into that fluid will probably be a very, very efficient route for treating patients.”

Treating brain disorders with cytokines

Developing molecules to infuse into the blood or CSF to communicate with the brain requires a better understanding of how cytokines affect neurons, astrocytes, and micro-glia over the course of a lifetime. Interferon gamma, for example, appears to have two faces when it comes to influencing neuronal circuits. In Kipnis’s studies of young mice, the cytokine was essential for the animals to be social. But an analysis of the brains of old mice shows that the same cytokine might be detrimental to making new neurons in aged mice. Giving the old mice an antibody that neutralizes the immune cytokine restored neurogenesis in the animals’ brains, a team of researchers reported in 2019.

A similar tactic might provide a novel way to treat various neuropsychiatric disorders such as schizophrenia. Analyses of immune cells in the blood of schizophrenia patients show that these individuals have higher levels of a variety of cytokines, including IL-13 and interferon gamma, than healthy individuals do. People with schizophrenia treated with anti-inflammatory and antipsychotic drugs also tend to have fewer cognitive problems than individuals treated with only antipsychotics, hinting that reducing cytokine levels could improve patients’ symptoms. While the neuronal changes that cause schizophrenia are far from clear, studies suggest that when certain neurons produce lower-than-expected levels of dopamine, they alert T cells to a problem, and the T cells respond by releasing cytokines that prompt disease-related deficits in memory, learning, social behavior, and resilience to stress.

Filiano and Kipnis have found evidence that a similar approach might work for helping individuals with autism. In experiments with mice lacking T cells, the researchers found that the animals not only had social deficits but also showed hyperactivity in neural circuits that often have abnormal activity in the brains of people with autism. Not only did social behavior improve when the team infused the mice with immune cells, but the animals’ abnormal neural activity subsided too. Meticulously tweaking the immune system might reverse the cognitive and social deficits of the disorder, the experiments suggest.

For now, the results leave Filiano wanting to know more. He explains, “We’re really interested in how these immune cells talk to the brain, how these signals get from the immune cells to these neural circuits, how that communication happens in health and disease.” 

Communication Signals

A growing number of knock-out experiments in mice have revealed several T cell–derived cytokines that influence learning, memory, and sociability.




Interferon gamma



Learning, long-term memory


Learning, long-term memory


Short-term memory, anxiety

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?