MIT scientists find a way to rejuvenate the immune system as we age

 

As people get older, the immune system often becomes less effective. Populations of T cells shrink, and the remaining cells may respond more slowly to germs. That slowdown can leave older adults more vulnerable to many kinds of infections.

To address this age related decline, scientists from MIT and the Broad Institute developed a method to temporarily reprogram liver cells in a way that strengthens T cell performance. The goal is to make up for the reduced output of the thymus, the organ where T cells normally mature.

In the study, the team used mRNA to deliver three important factors that support T cell survival. With this approach, they were able to rejuvenate the immune systems of mice. Older mice that received the treatment produced larger and more varied T cell populations after vaccination, and they also showed improved responses to cancer immunotherapy.

The researchers say that if this strategy can be adapted for patients, it could help people stay healthier as they age.

"If we can restore something essential like the immune system, hopefully we can help people stay free of disease for a longer span of their life," says Feng Zhang, the James and Patricia Poitras Professor of Neuroscience at MIT, who has joint appointments in the departments of Brain and Cognitive Sciences and Biological Engineering.

Zhang is also an investigator at the McGovern Institute for Brain Research at MIT, a core institute member at the Broad Institute of MIT and Harvard, and an investigator in the Howard Hughes Medical Institute. He is the senior author of the new study. Former MIT postdoc Mirco Friedrich is the lead author of the paper, which was published in Nature.

The Thymus and Why T Cells Decline With Age

The thymus is a small organ located in front of the heart, and it is essential for building a healthy supply of T cells. Inside the thymus, immature T cells go through a checkpoint process that helps create a diverse set of T cells. The thymus also releases cytokines and growth factors that help T cells survive.

But beginning in early adulthood, the thymus starts to shrink. This process is called thymic involution, and it reduces the body's ability to produce new T cells. By about age 75, the thymus is essentially nonfunctional.

"As we get older, the immune system begins to decline. We wanted to think about how can we maintain this kind of immune protection for a longer period of time, and that's what led us to think about what we can do to boost immunity," Friedrich says.

Earlier efforts to rejuvenate the immune system have often focused on sending T cell growth factors through the bloodstream, but that approach can cause harmful side effects. Other researchers are investigating whether transplanted stem cells could help regrow functional thymus tissue.

A Temporary Liver Factory Powered by mRNA

The MIT team chose a different strategy. They asked whether the body could be prompted to create a temporary "factory" that produces the same T cell stimulating signals typically made by the thymus.

"Our approach is more of a synthetic approach," Zhang says. "We're engineering the body to mimic thymic factor secretion."

They selected the liver for the job for several reasons. The liver can produce large amounts of protein even in old age. It is also easier to deliver mRNA to the liver than to many other organs. In addition, all circulating blood flows through the liver, including T cells, making it a practical place to release immune supporting signals into the bloodstream.

To build this factory, the researchers picked three immune cues involved in T cell maturation. They encoded these factors into mRNA and packaged the sequences into lipid nanoparticles. After injection into the bloodstream, the nanoparticles collect in the liver. Hepatocytes take up the mRNA and begin making the proteins encoded by it.

The three factors delivered were DLL1, FLT-3, and IL-7. These signals help immature progenitor T cells develop into fully differentiated T cells.

Vaccine and Cancer Immunotherapy Benefits in Older Mice

Experiments in mice showed multiple positive outcomes. In one test, the researchers injected the mRNA particles into 18 month old mice, roughly comparable to humans in their 50s. Because mRNA does not last long in the body, the team gave repeated doses over four weeks to keep the liver producing the factors consistently.

After the treatment, T cell populations increased substantially in both size and function.

The team then examined whether the approach improved vaccine responses. They vaccinated mice with ovalbumin, a protein found in egg whites that is often used to study immune reactions to a specific antigen. In 18 month old mice that received the mRNA treatment before vaccination, the number of cytotoxic T cells targeting ovalbumin doubled compared with untreated mice of the same age.

The researchers also found that the mRNA method could strengthen responses to cancer immunotherapy. They treated 18 month old mice with the mRNA, implanted tumors, and then gave the mice a checkpoint inhibitor drug. This drug targets PD-L1 and is intended to release the immune system's brakes so T cells can attack tumor cells more effectively.

Mice that received both the checkpoint inhibitor and the mRNA treatment had much higher survival rates and lived longer than mice that received the checkpoint inhibitor drug without the mRNA treatment.

The researchers determined that all three factors were required for the immune improvement. No single factor could reproduce the full effect. Next, the team plans to test the approach in additional animal models and search for other signaling factors that might further strengthen immune function. They also want to investigate how the treatment influences other immune cells, including B cells.

Other authors of the paper include Julie Pham, Jiakun Tian, Hongyu Chen, Jiahao Huang, Niklas Kehl, Sophia Liu, Blake Lash, Fei Chen, Xiao Wang, and Rhiannon Macrae.

The research was funded in part by the Howard Hughes Medical Institute, the K. Lisa Yang Brain-Body Center at MIT, Broad Institute Programmable Therapeutics Gift Donors, the Pershing Square Foundation, the Phillips family, J. and P. Poitras, and an EMBO Postdoctoral Fellowship.

Journal Reference:

1. Mirco J. Friedrich, Julie Pham, Jiakun Tian, Hongyu Chen, Jiahao Huang, Niklas Kehl, Sophia Liu, Blake Lash, Fei Chen, Xiao Wang, Rhiannon K. Macrae, Feng Zhang. Transient hepatic reconstitution of trophic factors enhances aged immunity. Nature, 2025; DOI: 10.1038/s41586-025-09873-4 

Courtesy:
Massachusetts Institute of Technology. "MIT scientists find a way to rejuvenate the immune system as we age." ScienceDaily. ScienceDaily, 29 December 2025. <www.sciencedaily.com/releases/2025/12/251227082718.htm>. 

 

 

 

 

The brain has a hidden language and scientists just found it

 

Scientists have developed a protein that can record the chemical messages brain cells receive, rather than focusing only on the signals they send out. These incoming signals are created when neurons release glutamate, a neurotransmitter that plays a vital role in brain communication. Although glutamate is essential for processes like learning and memory, its activity has been extremely difficult to measure because the signals are faint and happen very quickly.

This new tool makes it possible to detect those subtle chemical messages as they arrive, giving researchers access to a part of brain communication that has long been hidden.

Being able to observe incoming signals allows scientists to study how neurons process information. Each brain cell receives thousands of inputs, and how it combines those signals determines whether it produces an output. This process is thought to underlie decisions, thoughts, and memories, and studying it directly could help explain how the brain performs complex computations.

The advance also opens new paths for disease research. Problems with glutamate signaling have been linked to conditions such as Alzheimer's disease, schizophrenia, autism, epilepsy, and others. By measuring these signals more precisely, researchers may be able to identify the biological roots of these disorders.

Drug development could also benefit. Pharmaceutical companies can use these sensors to see how experimental treatments affect real synaptic activity, which may help speed up the search for more effective therapies.

Introducing a powerful glutamate sensor

The protein was engineered by researchers at the Allen Institute and HHMI's Janelia Research Campus. Known as iGluSnFR4 (pronounced 'glue sniffer'), it acts as a molecular "glutamate indicator." Its sensitivity allows it to detect even the weakest incoming signals exchanged between neurons.

By revealing when and where glutamate is released, iGluSnFR4 provides a new way to interpret the complex patterns of brain activity that support learning, memory, and emotion. It gives scientists the ability to watch neurons communicate inside the brain in real time. The findings were recently published in Nature Methods and could significantly change how neural activity is measured and analyzed in neuroscience research.

How brain cells communicate

To understand the impact of this advance, it helps to look at how neurons interact. The brain contains billions of neurons that communicate by sending electrical signals along branch-like structures called axons. When an electrical signal reaches the end of an axon, it cannot cross the small gap to the next neuron, which is known as a synapse.

Instead, the signal triggers the release of neurotransmitters into the synapse. Glutamate is the most common of these chemical messengers and plays a key role in memory, learning, and emotion. When glutamate reaches the next neuron, it can cause that cell to fire, continuing the chain of communication.

From fragments to the full conversation

This process can be compared to falling dominos, but it is far more complex. Each neuron receives input from thousands of others, and only certain combinations and patterns of activity will trigger the receiving neuron to fire. With this new protein sensor, scientists can now identify which patterns of incoming activity lead to that response.

Until now, observing these incoming signals in living brain tissue was nearly impossible. Previous technologies were too slow or lacked the sensitivity needed to measure activity at individual synapses. As a result, researchers could only see pieces of the communication process rather than the full exchange. This new approach allows them to capture the entire conversation.

Making sense of neural connections

"It's like reading a book with all the words scrambled and not understanding the order of the words or how they're arranged," said Kaspar Podgorski, Ph.D., a lead author of the study and senior scientist at the Allen Institute. "I feel like what we're doing here is adding the connections between those neurons and by doing that, we now understand the order of the words on the pages, and what they mean."

Before protein sensors like iGluSnFR4 were available, researchers could only measure outgoing signals from neurons. This left a major gap in understanding, since the incoming signals were too fast and too faint to detect.

"Neuroscientists have pretty good ways of measuring structural connections between neurons, and in separate experiments, we can measure what some of the neurons in the brain are saying, but we haven't been good at combining these two kinds of information. It's hard to measure what neurons are saying to which other neurons," Podgorski said. "What we have invented here is a way of measuring information that comes into neurons from different sources, and that's been a critical part missing from neuroscience research."

Collaboration behind the breakthrough

"The success of iGluSnFR4 stems from our close collaboration started at HHMI's Janelia Research Campus between the GENIE Project team and Kaspar's lab. That research has extended to the phenomenal in vivo characterization work done by the Allen Institute's Neural Dynamics group," said Jeremy Hasseman, Ph.D., a scientist with HHMI's Janelia Research Campus. "This was a great example of collaboration across labs and institutes to enable new discoveries in neuroscience."

A new window into brain function

This discovery overcomes a major limitation in modern neuroscience by making it possible to directly observe how neurons receive information. With iGluSnFR4 now available to researchers through Addgene, scientists have a powerful new tool to explore brain function in greater detail. As this technology spreads, it may help reveal answers to some of the brain's most enduring questions.

Journal Reference:

1. Abhi Aggarwal, Adrian Negrean, Yang Chen, Rishyashring Iyer, Daniel Reep, Anyi Liu, Anirudh Palutla, Michael E. Xie, Bryan J. MacLennan, Kenta M. Hagihara, Lucas W. Kinsey, Julianna L. Sun, Pantong Yao, Jihong Zheng, Arthur Tsang, Getahun Tsegaye, Yonghai Zhang, Ronak H. Patel, Benjamin J. Arthur, Julien Hiblot, Philipp Leippe, Miroslaw Tarnawski, Jonathan S. Marvin, Jason D. Vevea, Srinivas C. Turaga, Alison G. Tebo, Matteo Carandini, L. Federico Rossi, David Kleinfeld, Arthur Konnerth, Karel Svoboda, Glenn C. Turner, Jeremy P. Hasseman, Kaspar Podgorski. Glutamate indicators with increased sensitivity and tailored deactivation rates. Nature Methods, 2025; DOI: 10.1038/s41592-025-02965-z 

Courtesy:

Allen Institute. "The brain has a hidden language and scientists just found it." ScienceDaily. ScienceDaily, 29 December 2025. <www.sciencedaily.com/releases/2025/12/251225235950.htm>.

 

 

 

 

 

 

Ramanujan’s 100-year-old pi formula is still revealing the Universe

 

Most people first encounter the irrational number π (pi) -- commonly approximated as 3.14 and extending infinitely without repeating -- during school lessons about circles. In recent decades, advances in computing have pushed this familiar constant far beyond the classroom, with powerful supercomputers now calculating pi to trillions of decimal places.

Researchers have now uncovered an unexpected twist. Physicists at the Centre for High Energy Physics (CHEP), Indian Institute of Science (IISc) report that mathematical formulas developed a century ago to compute pi are closely linked to some of today's most important ideas in fundamental physics. These connections appear in theoretical descriptions of percolation, fluid turbulence, and even certain features of black holes.

Ramanujan's Remarkable Pi Formulae

In 1914, shortly before leaving Madras for Cambridge, renowned Indian mathematician Srinivasa Ramanujan published a paper presenting 17 different formulas for calculating pi. These expressions were strikingly efficient, allowing pi to be computed much faster than existing techniques of the time. Despite containing only a small number of mathematical terms, the formulas produced an impressive number of accurate digits.

Their impact has endured. Ramanujan's methods became foundational to modern mathematical and computational approaches for calculating pi, including those used by today's most advanced machines. "Scientists have computed pi up to 200 trillion digits using an algorithm called the Chudnovsky algorithm," says Aninda Sinha, Professor at CHEP and senior author of the study. "These algorithms are actually based on Ramanujan's work."

A Deeper Question Behind the Mathematics

For Sinha and Faizan Bhat, the study's first author and a former IISc PhD student, the mystery went beyond computational efficiency. They asked why such powerful formulas should exist in the first place. Rather than treating them as purely abstract results, the team searched for an explanation rooted in physics.

"We wanted to see whether the starting point of his formulae fit naturally into some physics," says Sinha. "In other words, is there a physical world where Ramanujan's mathematics appears on its own?"

Where Pi Meets Scale Invariance and Physics Extremes

Their investigation led them to a broad family of theories known as conformal field theories, and more specifically to logarithmic conformal field theories. These theories describe systems that exhibit scale invariance symmetry -- meaning they look the same regardless of how closely they are examined, similar to fractals.

A familiar physical example appears at the critical point of water, defined by a precise temperature and pressure at which liquid water and water vapor become indistinguishable. At this point, water displays scale invariance symmetry, and its behavior can be captured using conformal field theory. Similar critical behavior arises in percolation (how substances spread through a material), during the onset of turbulence in fluids, and in certain theoretical treatments of black holes. These phenomena fall within the domain of logarithmic conformal field theories.

Using Ramanujan's Structure to Solve Physics Problems

The researchers discovered that the mathematical framework at the heart of Ramanujan's pi formulas also appears in the equations underlying these logarithmic conformal field theories. By exploiting this shared structure, they were able to compute key quantities within the theories more efficiently. Such calculations could ultimately improve scientists' understanding of complex processes like turbulence and percolation.

The approach mirrors Ramanujan's own method of starting from a compact mathematical expression and rapidly arriving at precise results for pi. "[In] any piece of beautiful mathematics, you almost always find that there is a physical system which actually mirrors the mathematics," says Bhat. "Ramanujan's motivation might have been very mathematical, but without his knowledge, he was also studying black holes, turbulence, percolation, all sorts of things."

A Century-Old Insight With Modern Impact

The findings reveal that Ramanujan's formulas, developed more than 100 years ago, offer a previously hidden advantage for making modern high-energy physics calculations faster and more manageable. Beyond their practical value, the researchers say the work highlights the extraordinary reach of Ramanujan's ideas.

"We were simply fascinated by the way a genius working in early 20th century India, with almost no contact with modern physics, anticipated structures that are now central to our understanding of the universe," says Sinha.

Journal Reference:

1. Faizan Bhat, Aninda Sinha. Ramanujan’s 1/π Series and Conformal Field Theories. Physical Review Letters, 2025; 135 (23) DOI: 10.1103/c38g-fd2v 

Courtesy: Indian Institute of Science (IISc). "Ramanujan’s 100-year-old pi formula is still revealing the Universe." ScienceDaily. ScienceDaily, 16 December 2025. <www.sciencedaily.com/releases/2025/12/251216081949.htm>. 

 

 

 

A flesh-eating fly once eradicated is moving back toward the U.S.

When the New World screwworm last spread across the United States, it caused widespread damage to livestock and took decades to eliminate. That history is now driving a new effort by researchers at the University of California Riverside, who are working to stop the parasitic fly before it can reestablish itself.

Despite its name, the New World screwworm is not a worm. It is the larval, or maggot, stage of a metallic-looking blowfly known as Cochliomyia hominivorax. Most blowflies are harmless and play an important role by breaking down dead animals. This species is different because it feeds on living tissue.

"Not all blowflies are this species. We don't have to be afraid of all flies," said Amy Murillo, UCR assistant professor of entomology and principal investigator of the project. "But this particular species isn't one we want here."

Monitoring California for Early Warning Signs

Backed by $507,000 from the California Department of Food and Agriculture, UCR entomologists are launching a statewide monitoring program to detect any early signs of the fly's return. The screwworm lays its eggs in open wounds on warm-blooded animals, including humans. Once the eggs hatch, the maggots burrow into flesh in a corkscrew motion, which is how the insect got its name.

The New World screwworm was once common across California and much of the southern United States. About 60 years ago, it was successfully wiped out using a large-scale program that released millions of sterile male flies. Because female screwworms mate only once, the strategy caused the population to collapse. That effort pushed the species south to Panama, where the U.S. Department of Agriculture has maintained a protective barrier ever since.

A Growing Threat to the North

In recent years, the screwworm has started appearing again in parts of Central America and southern Mexico. Experts believe the resurgence is linked to the movement of infested animals, often transported by people who unknowingly help the pest spread.

"It hasn't been found in California yet, but it's within 70 miles of the Texas border," Murillo said. "We need to be prepared."

Traps, Lures, and Early Detection

Murillo is working with Alec Gerry, a UCR veterinary entomology professor and CE Specialist, to deploy traps across California. These traps use a lure developed by the USDA that mimics the smell of rotting flesh. While many blowfly species are attracted to it, the lure is also effective at drawing in the New World screwworm. Researchers will regularly check the traps for any sign that the invasive fly has crossed into the state.

The project also includes an outreach effort aimed at veterinarians, livestock handlers, and entomologists. These groups are the most likely to encounter screwworm infestations early, and catching an outbreak quickly is key to stopping it from spreading.

Why Livestock Face the Greatest Risk

Screwworms pose a serious threat to farm animals, especially those with common injuries from barbed wire, birthing, or procedures such as dehorning. Without treatment, infestations can quickly worsen as flies lay more eggs, enlarging wounds and sometimes leading to death.

California's dairy and cattle industries are particularly vulnerable. "Most people think of citrus or avocados as being our top exports, but it's actually dairy that leads our agricultural economy," Murillo said.

Public Awareness as a Line of Defense

Even though the fly has not been detected in California, researchers say public awareness is an important part of prevention. "Not all blowflies are harmful, and many are beneficial," Murillo said. "But if you notice something unusual on your pet or livestock, reach out to a vet. Don't ignore it."

If an outbreak were to happen again, Murillo said the same sterile insect technique used in the past would still be the most effective tool for stopping it.

Rare but Possible Human Infections

Although screwworms primarily target animals, humans can also be affected under certain conditions, particularly when open wounds go untreated. Infections have been documented in people who traveled to regions where the fly is active. Even so, livestock remain the primary concern.

"They need an opening in the flesh, and it doesn't have to be large. They don't make wounds, but they do exploit them," Murillo said.

Staying Alert Without Panic

Murillo emphasized that the goal is preparedness, not fear.

"You don't have to worry that they're going to start eating your flesh," Murillo said. "But we do hope that this project will help people to be more vigilant in recognizing and preventing screwworm infestations in animals should they return to California."

Story Source:

Materials provided by University of California - Riverside. Note: Content may be edited for style and length.

 

Courtesy:

University of California - Riverside. "A flesh-eating fly once eradicated is moving back toward the U.S.." ScienceDaily. ScienceDaily, 18 December 2025. <www.sciencedaily.com/releases/2025/12/251217082501.htm>.  

AI detects cancer but it’s also reading who you are

 

At a glance:

• A new study shows that artificial intelligence systems used to diagnose cancer from pathology slides do not perform equally for all patients, with accuracy varying across different demographic groups.
• Researchers pinpointed three key reasons behind this bias and created a new approach that significantly reduced these differences.
• The results emphasize why medical AI must be routinely evaluated for bias to help ensure fair and reliable cancer care for everyone.

Pathology and the Foundations of Cancer Diagnosis

For decades, pathology has been essential to how doctors diagnose and treat cancer. A pathologist studies an extremely thin slice of human tissue under a microscope, searching for visual signs that reveal whether cancer is present and, if so, what type and stage it has reached.

To a trained specialist, examining a pink, swirling tissue sample dotted with purple cells is like grading a test without a name on it -- the slide contains vital information about the disease, but it offers no clues about who the patient is.

When AI Sees More Than Expected

That assumption does not fully apply to artificial intelligence systems now entering pathology labs. A new study led by researchers at Harvard Medical School shows that pathology AI models can infer demographic details directly from tissue slides. This unexpected ability can introduce bias into cancer diagnosis across different patient groups.

After evaluating several widely used AI models designed to identify cancer, the researchers found that these systems did not perform equally for all patients. Diagnostic accuracy varied based on patients' self-reported race, gender, and age. The team also uncovered several reasons why these disparities occur.

To address the issue, the researchers developed a framework called FAIR-Path, which significantly reduced bias in the tested models.

"Reading demographics from a pathology slide is thought of as a 'mission impossible' for a human pathologist, so the bias in pathology AI was a surprise to us," said senior author Kun-Hsing Yu, associate professor of biomedical informatics in the Blavatnik Institute at HMS and HMS assistant professor of pathology at Brigham and Women's Hospital.

Yu emphasized that recognizing and correcting bias in medical AI is critical, since it can directly influence diagnostic accuracy and patient outcomes. The success of FAIR-Path suggests that improving fairness in cancer pathology AI, and possibly other medical AI tools, may not require major changes to existing systems.

The work, which was supported in part by federal funding, is described Dec. 16 in Cell Reports Medicine.

Putting Cancer AI to the Test

Yu and his colleagues examined bias in four commonly used pathology AI models currently being developed for cancer diagnosis. These deep-learning systems were trained on large collections of labeled pathology slides, allowing them to learn biological patterns and apply that knowledge to new samples.

The team evaluated the models using a large, multi-institutional dataset that included pathology slides from 20 different types of cancer.

Across all four models, performance gaps consistently emerged. The AI systems were less accurate for certain demographic groups defined by race, gender, and age. For example, the models struggled to distinguish lung cancer subtypes in African American patients and in male patients. They also showed reduced accuracy when classifying breast cancer subtypes in younger patients. In addition, the models had difficulty detecting breast, renal, thyroid, and stomach cancers in some demographic groups. Overall, these disparities appeared in roughly 29 percent of the diagnostic tasks analyzed.

According to Yu, these errors arise because the AI systems extract demographic information from the tissue images -- and then rely on patterns linked to those demographics when making diagnostic decisions.

The findings were unexpected. "Because we would expect pathology evaluation to be objective," Yu said. "When evaluating images, we don't necessarily need to know a patient's demographics to make a diagnosis."

This led the researchers to ask a key question: Why was pathology AI failing to meet the same standard of objectivity?

Why Bias Appears in Pathology AI

The team identified three main contributors to the bias.

First, training data are often uneven. Tissue samples are easier to obtain from some demographic groups than others, resulting in imbalanced datasets. This makes it harder for AI models to accurately diagnose cancers in groups that are underrepresented, including some populations defined by race, age, or gender.

However, Yu noted that "the problem turned out to be much deeper than that." In several cases, the models performed worse for certain demographic groups even when sample sizes were similar.

Further analysis pointed to differences in disease incidence. Some cancers occur more frequently in specific populations, allowing AI models to become especially accurate for those groups. As a result, the same models may struggle to diagnose cancers in populations where those diseases are less common.

The researchers also found that AI models can detect subtle molecular differences across demographic groups. For example, the systems may identify mutations in cancer driver genes and use them as shortcuts to classify cancer type -- which can reduce accuracy in populations where those mutations are less prevalent.

"We found that because AI is so powerful, it can differentiate many obscure biological signals that cannot be detected by standard human evaluation," Yu said.

Over time, this can cause AI models to focus on signals tied more closely to demographics than to the disease itself, weakening diagnostic performance across diverse patient groups.

Taken together, Yu said, these findings show that bias in pathology AI is influenced not only by the quality and balance of training data, but also by the way the models are trained to interpret what they see.

A New Approach to Reducing Bias

After identifying the sources of bias, the researchers set out to correct them.

They developed FAIR-Path, a framework based on an existing machine-learning method known as contrastive learning. This approach modifies AI training so that models focus more strongly on critical distinctions, such as differences between cancer types, while reducing attention to less relevant differences, including demographic characteristics.

When FAIR-Path was applied to the tested models, diagnostic disparities dropped by about 88 percent.

"We show that by making this small adjustment, the models can learn robust features that make them more generalizable and fairer across different populations," Yu said.

The result is encouraging, he added, because it suggests that meaningful reductions in bias are possible even without perfectly balanced or fully representative training datasets.

Looking ahead, Yu and his team are working with institutions worldwide to study pathology AI bias in regions with different demographics, clinical practices, and laboratory settings. They are also exploring how FAIR-Path could be adapted for situations with limited data. Another area of interest is understanding how AI-driven bias contributes to broader disparities in health care and patient outcomes.

Ultimately, Yu said, the goal is to develop pathology AI systems that support human experts by delivering fast, accurate, and fair diagnoses for all patients.

"I think there's hope that if we are more aware of and careful about how we design AI systems, we can build models that perform well in every population," he said.

Authorship, funding, disclosures

Additional authors on the study include Shih-Yen Lin, Pei-Chen Tsai, Fang-Yi Su, Chun-Yen Chen, Fuchen Li, Junhan Zhao, Yuk Yeung Ho, Tsung-Lu Michael Lee, Elizabeth Healey, Po-Jen Lin, Ting-Wan Kao, Dmytro Vremenko, Thomas Roetzer-Pejrimovsky, Lynette Sholl, Deborah Dillon, Nancy U. Lin, David Meredith, Keith L. Ligon, Ying-Chun Lo, Nipon Chaisuriya, David J. Cook, Adelheid Woehrer, Jeffrey Meyerhardt, Shuji Ogino, MacLean P. Nasrallah, Jeffrey A. Golden, Sabina Signoretti, and Jung-Hsien Chiang.

Funding was provided by the National Institute of General Medical Sciences and the National Heart, Lung, and Blood Institute at the National Institutes of Health (grants R35GM142879, R01HL174679), the Department of Defense (Peer Reviewed Cancer Research Program Career Development Award HT9425-231-0523), the American Cancer Society (Research Scholar Grant RSG-24-1253761-01-ESED), a Google Research Scholar Award, a Harvard Medical School Dean's Innovation Award, the National Science and Technology Council of Taiwan (grants NSTC 113-2917-I-006-009, 112-2634-F-006-003, 113-2321-B-006-023, 114-2917-I-006-016), and a doctoral student scholarship from the Xin Miao Education Foundation.

Ligon was a consultant of Travera, Bristol Myers Squibb, Servier, IntegraGen, L.E.K. Consulting, and Blaze Bioscience; received equity from Travera; and has research funding from Bristol Myers Squibb and Lilly. Vremenko is a cofounder and shareholder of Vectorly.

The authors prepared the initial manuscript and used ChatGPT to edit selected sections to improve readability. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Journal Reference:

1. Shih-Yen Lin, Pei-Chen Tsai, Fang-Yi Su, Chun-Yen Chen, Fuchen Li, Junhan Zhao, Yuk Yeung Ho, Tsung-Lu Michael Lee, Elizabeth Healey, Po-Jen Lin, Ting-Wan Kao, Dmytro Vremenko, Thomas Roetzer-Pejrimovsky, Lynette Sholl, Deborah Dillon, Nancy U. Lin, David Meredith, Keith L. Ligon, Ying-Chun Lo, Nipon Chaisuriya, David J. Cook, Adelheid Woehrer, Jeffrey Meyerhardt, Shuji Ogino, MacLean P. Nasrallah, Jeffrey A. Golden, Sabina Signoretti, Jung-Hsien Chiang, Kun-Hsing Yu. Contrastive learning enhances fairness in pathology artificial intelligence systems. Cell Reports Medicine, 2025; 6 (12): 102527 DOI: 10.1016/j.xcrm.2025.102527 

Courtesy:

Harvard Medical School. "AI detects cancer but it’s also reading who you are." ScienceDaily. ScienceDaily, 17 December 2025. <www.sciencedaily.com/releases/2025/12/251217231230.htm>. 

 

 

 

 

 

 

A hidden T cell switch could make cancer immunotherapy work for more people

Over the past ten years, T cell immunotherapy has emerged as one of the most promising developments in cancer treatment. These therapies work by training a patient's own immune system to detect and destroy dangerous cells. Despite their success, scientists have struggled to fully explain how these treatments function at a molecular level. This lack of understanding has slowed progress, especially since T cell therapies work well for only a small number of cancer types and fail in most others, for reasons that have remained unclear. Gaining insight into their modus operandi could help make these therapies effective for far more patients.

Scientists at The Rockefeller University have now uncovered crucial details about the T cell receptor (TCR), a protein complex embedded in the cell membrane that plays a central role in T cell therapies. Using cryo-EM, researchers from the Laboratory of Molecular Electron Microscopy studied the receptor in a biochemical setting designed to closely resemble its native milieu. They discovered that the TCR behaves like a jack-in-the-box, staying compact until it encounters an antigen or another suspicious particle, at which point it rapidly opens. This behavior contradicts what earlier cryo-EM studies of the receptor had shown.

The findings, published in Nature Communications, could help researchers improve and expand the use of T cell immunotherapies.

"This new fundamental understanding of how the signaling system works may help re-engineer that next generation of treatments," says first author Ryan Notti, an instructor in clinical investigation in Walz's lab and a special fellow in the Department of Medicine at Memorial Sloan Kettering Cancer Center, where he treats patients with sarcomas, or cancers that arise in soft tissue or bone.

"The T cell receptor is really the basis of virtually all oncological immunotherapies, so it's remarkable that we use the system but really have had no idea how it actually works -- and that's where basic science steps in," says Walz, a world expert in cryo-EM imaging. "This is some of the most important work to ever come out of my lab."

How T Cells Detect Threats

Walz's lab focuses on producing detailed images of macromolecular complexes, especially proteins found in cell membranes that help cells communicate with their surroundings. The TCR is one such complex. Made up of multiple proteins, it enables T cells to recognize antigens displayed by human leukocyte antigen (HLA) complexes on other cells. This recognition process is what T cell therapies rely on to mobilize the immune system against cancer.

Although scientists have known the individual parts of the TCR for many years, the earliest steps that trigger its activation have remained elusive. Notti, who works as both a physician and a researcher, found this gap especially troubling because many of his sarcoma patients were not benefiting from T cell immunotherapies.

"Determining that would help us understand how the information gets from outside the cell, where those antigens are being presented by HLAs, to the inside of the cell, where signaling turns on the T cell," he says.

Notti earned his Ph.D. in structural microbiology at Rockefeller before moving into oncology, and he suggested to Walz that they investigate this unanswered question together.

Rebuilding the TCR's Natural Environment

Walz's team is known for creating custom membrane environments that closely mimic the natural surroundings of membrane proteins. "We can change the biochemical composition, the thickness of the membrane, the tension and curvature, the size -- all kinds of parameters that we know have an influence on the embedded protein," Walz says.

For this study, the researchers set out to observe the TCR in conditions that closely resemble those inside a living cell. They placed the receptor into a nanodisc, a tiny disc-shaped section of membrane held in solution by a scaffold protein wrapped around its edge. Assembling the full receptor was difficult, and "getting all eight of these proteins properly assembled into the nanodisc was challenging," Notti says.

Previous structural studies of the TCR had relied on detergent, which often strips away the surrounding membrane. Walz notes that this was the first time the receptor complex had been restored to a membrane environment for detailed imaging.

Seeing the Receptor Switch On

Once the TCR was embedded in the nanodisc, the researchers used cryo-EM to visualize it. The images showed that the receptor remains closed and compact when inactive. When it encounters an antigen-presenting molecule, however, the structure opens and extends outward, resembling a wide-reaching motion.

The result surprised the team. "The data that were available when we began this research depicted this complex as being open and extended in its dormant state," Notti explains. "As far as anyone knew, the T cell receptor didn't undergo any conformational changes when binding to these antigens. But we found that it does, springing open like a sort of jack-in-the-box."

The researchers believe two factors made this discovery possible. First, they carefully recreated the TCR's in vivo membrane environment using the right lipid mixture. Second, they reinserted the receptor into a membrane using nanodiscs before conducting cryo-EM imaging. They found that an intact membrane keeps the receptor in a closed position until activation occurs. In earlier studies, detergent may have removed this restraint, allowing the receptor to open prematurely.

"It was important that we used a lipid mixture that resembled that of the native T cell membrane," says Walz. "If we had just used a model lipid, we wouldn't have seen this closed dormant state either."

Implications for Cancer Therapies and Vaccines

The team believes their findings could help improve treatments that rely on T cell receptors. "Re-engineering the next generation of immunotherapies tops the charts in terms of unmet clinical needs," Notti says. "For example, adoptive T cell therapies are being used successfully to treat certain very rare sarcomas, so one could imagine using our insights to re-engineer the sensitivity of those receptors by tuning their activation threshold."

Walz also sees potential applications beyond cancer therapy. "This information may be used for vaccine design as well," he adds. "People in the field can now use our structures to see refined details about the interactions between different antigens presented by HLA and T cell receptors. Those different modes of interaction might have some implication for how the receptor functions -- and ways to optimize it."

Journal Reference:

1. Ryan Q. Notti, Fei Yi, Søren Heissel, Martin W. Bush, Zaki Molvi, Pujita Das, Henrik Molina, Christopher A. Klebanoff, Thomas Walz. The resting and ligand-bound states of the membrane-embedded human T-cell receptor–CD3 complex. Nature Communications, 2025; 16 (1) DOI: 10.1038/s41467-025-66939-7 

Courtesy: Rockefeller University. "A hidden T cell switch could make cancer immunotherapy work for more people." ScienceDaily. ScienceDaily, 18 December 2025. <www.sciencedaily.com/releases/2025/12/251218074429.htm>.