The breakthrough comes from a team led by scientists at the Johns Hopkins Kimmel Cancer Center, its Bloomberg~Kimmel Institute for Cancer Immunotherapy, and the Johns Hopkins University School of Medicine. Their findings, published in Nature, identify a protein called claudin-4 as the essential docking point for a bacterial toxin known as BFT.

Even more promising, the researchers have already used this discovery to develop a molecular decoy that intercepts the toxin before it can damage colon cells. The strategy successfully protected mice from the toxin’s harmful effects, suggesting a possible new approach to preventing bacterial damage associated with colorectal cancer. The research received partial support from the National Institutes of Health.

“We’ve made several attempts over time to identify the receptor, so this is an exciting moment,” says senior author Cynthia Sears, M.D., Bloomberg~Kimmel Professor of Cancer Immunotherapy and professor of medicine at Johns Hopkins. “Understanding how bacterial toxins work can open doors to new approaches for detection and therapy for associated diseases, including diarrhea, colorectal cancer and bloodstream infections.”

How a Common Gut Bacterium Can Promote Colon Cancer

The investigation centers on Bacteroides fragilis, a bacterium that can be found in the intestines of up to 20% of healthy people. Although the presence of this bacterium does not automatically mean someone will develop cancer, certain strains produce a powerful toxin capable of disrupting the colon’s protective lining, causing inflammation, and encouraging tumor growth.

The toxin, known as B. fragilis toxin or BFT, has been under investigation for years. In a landmark 2009 study published in Nature Medicine, Sears and her colleagues demonstrated that the toxin could promote the development of colon tumors. Their work helped establish an important connection between toxin-producing gut bacteria, chronic inflammation, and colorectal cancer.

Earlier experiments also revealed that BFT damages a protein called E-cadherin, which helps hold neighboring cells together and maintain the colon’s protective barrier. When this protein is broken apart, the barrier becomes compromised, potentially allowing harmful substances and bacteria to interfere with the surrounding tissue.

However, one critical question remained unanswered. Although BFT could break down E-cadherin, the toxin did not appear to attach directly to that protein. Researchers suspected that another molecule was helping the toxin reach its target, but repeated attempts to identify it had failed.

Scientists Identify the Missing Link Using CRISPR

To solve the mystery, the researchers turned to CRISPR, a powerful gene-editing technology that allows scientists to selectively disable genes and investigate what they do.

Maxwell White, an M.D./Ph.D. candidate working in Sears’ laboratory, led a genome-wide CRISPR screening effort in collaboration with Matthew Waldor’s laboratory at Harvard Medical School. The team systematically switched off genes in colon epithelial cells, the cells that form the inner lining of the colon, to determine which ones were necessary for the toxin to attach and cause damage.

One gene stood out dramatically. It produces claudin-4, a protein that helps neighboring cells form tight junctions, the seals that limit the movement of substances between cells and help maintain the intestinal barrier.

When the researchers disabled the gene responsible for claudin-4, BFT could no longer bind to the colon cells. Without that initial attachment, the toxin was unable to attack E-cadherin.

The finding revealed that claudin-4 acts as a receptor for the bacterial toxin. In effect, the toxin must first latch onto this protein before it can carry out its damaging activity.

“It took a while to get the assay working and validate the approach, but once we were able to do the screen, claudin-4 was a clear, resounding top hit,” says White. “That was an exciting moment.”

An Unexpected Discovery Reveals a New Bacterial Attack Strategy

The identity of the receptor surprised the scientists.

Sears and other researchers had expected BFT to interact with a protein involved in cellular signaling, such as a G protein-coupled receptor. These receptors act somewhat like communication switches, allowing cells to respond to signals from their surroundings.

Claudin-4, however, serves a very different purpose. Rather than functioning primarily as a signaling receptor, it helps maintain the physical connections between neighboring cells.

The discovery also revealed an unusual way for a bacterial toxin to operate. BFT belongs to a class of enzymes called proteases, which cut other proteins. Most proteases act directly on their targets, but BFT appears to require a separate receptor before it can reach and damage E-cadherin.

When the team reviewed previous scientific research, they could not identify another toxin known to use the same mechanism.

To establish that BFT and claudin-4 truly bind to each other, the Johns Hopkins researchers collaborated with structural biologists F. Xavier Gomis-Rüth and Ulrich Eckhard at the Molecular Biology Institute of Barcelona.

Using techniques that measure how molecules interact, White and the Barcelona team showed that BFT and claudin-4 form a tightly bound complex in laboratory experiments. Each toxin molecule paired with one receptor molecule, providing direct physical evidence of the connection.

A Clever Molecular Decoy Blocks the Dangerous Gut Toxin

Identifying the receptor did more than explain how the toxin works. It also gave the scientists an idea for stopping it.

If BFT needs claudin-4 to attach to colon cells, perhaps a substitute could lure the toxin away from its intended target.

To test that possibility, the team developed a molecular decoy. This was a soluble protein carrying sequences from claudin-4, designed to attract BFT without allowing it to attack the colon’s protective lining.

The researchers then worked with Min Dong’s laboratory at Harvard Medical School, including Kang Wang and colleagues, to investigate the approach in mice. These experiments allowed the scientists to examine how the toxin behaves in the more complicated environment of a living intestine.

The results were encouraging. Instead of binding to claudin-4 receptors on colon cells, BFT attached to the decoy proteins. By intercepting the toxin, the decoys successfully protected the mice from BFT-induced damage.

The finding suggests that it may be possible to interfere with the toxin’s first step rather than trying to repair the damage after it occurs.

“This approach could be iterated upon with small molecules or other biologics that have better pharmacological properties,” says White. The team is now exploring which molecular approaches might be most successful to block the toxin.

Small molecules are compounds that can sometimes be developed into medicines, while biologics are treatments made from or involving biological materials, such as proteins. Future versions of the decoy strategy could potentially be designed to work more effectively in the body or be easier to administer.

What This Discovery Could Mean for Colon Cancer Prevention

Although the findings identify a promising new target, important questions remain.

Scientists have established that BFT attaches to claudin-4 and that blocking this interaction can protect mice from toxin-induced damage. However, they have not yet captured the exact experimental three-dimensional structure showing how the two molecules fit together.

Even advanced artificial intelligence tools, including AlphaFold, have been unable to fully resolve the interaction. Understanding its precise structure could help researchers design more effective compounds that prevent the toxin from binding.

Further experiments will also be needed to determine whether the decoy approach can safely and effectively prevent toxin-related disease in humans. The mouse experiments demonstrated protection against tissue damage, but they did not establish that the strategy prevents colorectal cancer.

Nevertheless, the discovery resolves a major question that has persisted since the connection between BFT and colon tumor formation was established in 2009. By identifying the toxin’s previously unknown point of attachment, researchers have uncovered a potential way to interrupt a harmful biological process before it begins.

The findings could ultimately support new strategies for detecting or treating conditions associated with bacterial toxins, including colorectal cancer, diarrhea, and bloodstream infections.

Study Authors, Funding, and Disclosures

Additional researchers who contributed to the study include Jason Chen, Shaoguang Wu, Abby L. Geis, and Jessica Queen of Johns Hopkins, along with Hailong Zhang, Karthik Hullahalli, and Jie Zhang of Harvard Medical School.

Funding was provided by the Bloomberg~Kimmel Institute for Cancer Immunotherapy, Janssen Research and Development, Cancer Research UK, the National Institutes of Health (grant numbers R01 AI042347, R01 NS080833, R01 NS117626, R01 AI170835 and R01 AI189789) and the Howard Hughes Medical Institute.

Previous research by the Oued Beht Archaeological Project had already brought to light what is likely the earliest and largest agricultural complex ever discovered in Africa outside the Nile Valley. It dates to around 5,400 to 4,900 years ago. New excavations carried out in 2023 – in which we took part – have now deepened our understanding of the site.

Our new study shows how we uncovered the earliest known large-scale construction of a probably defensive nature in north Africa beyond the Nile Valley. It’s about 4,000 to 3,500 years old.

It’s not known exactly what threats these wall-like structures were designed to counter. They likely relate to increasing competition over land, resources, and possibly control of movement through the landscape, rather than to “invaders”. Constructions like these suggest communities felt the need to protect themselves or assert power over strategic locations.

This discovery pushes back the evidence for organized defensive architecture in the Maghreb and suggests that societies in this region at this time were more complex and structured than previously believed.

Our study also reveals how the ridge of Oued Beht would become a focal point again in the medieval period, during the 11th to 13th centuries AD. The same defensive structure was rebuilt, a hamlet or farm was established on a nearby hilltop, and people were buried in a cemetery between the two areas.

Together these findings provide rare archaeological insights into rural communities during both prehistoric and medieval times. They help us understand how societies developed and maintained themselves in a region too often left out of the history books.

A prehistoric hub in the Maghreb

For a long time, the later prehistory of north Africa’s Maghreb region (the Mediterranean countries of Morocco, Algeria, and Tunisia) has been poorly understood. Compared with neighboring areas such as Iberia in south-western Europe or Athens in the eastern Mediterranean, archaeological evidence was often fragmentary. In the last few years this has changed completely.

Excavations at Oued Beht have revealed a large concentration of human activity spread across at least ten hectares. The earliest major phase dates to the mid-fourth and early third millennia BC, or the Final Neolithic period.

Environmental evidence from plant and animal remains in the prehistoric era indicates a farming community cultivating crops such as barley and wheat. They also exploited wild plants and kept livestock.

The site was also an important center of craft production. Thousands of polished stone axes and other tools have been discovered, along with finely made pottery, sometimes decorated with painted patterns.

It’s still uncertain whether prehistoric Oued Beht functioned as a permanent village, a seasonal center, or something less familiar. But it was clearly a major focus in the landscape of north-west Morocco.

The defensive wall

The presence of potentially ancient earthworks and walls had been noted since the 1930s. The French archaeologist Armand Ruhlmann first documented a configuration of three walls, with the fourth side protected by the natural scarp of the ridge, describing a defensive éperon barré (barred spur). Their precise nature and date remained uncertain until now.

New excavations have clarified the existence of a multi-phase ditch, rampart and wall built across a narrow section of the Oued Beht ridge between the river and a ravine. This cut off and protected the ridge’s southern tip.

Radiocarbon dating of deposits associated with these structures gave us the timelines in our study.

They place its original construction at a time equivalent to the Bronze Age of the remaining western Mediterranean. This makes it the earliest confirmed defensive construction in north Africa outside Egypt.

What was this construction for?

This structure raises important questions about the society that built it. Building an embankment and ditch system required planning, coordination and the mobilization of labor. This suggests a community capable of organizing complex construction initiatives.

During the earlier Neolithic phase, activity at Oued Beht focused on the larger and more open northern area of the site. By the period of our new study, focus had shifted to the narrower, more defensible southern ridge. This may reflect a shift in settlement strategies.

Mediterranean connections

The second millennium BC was a dynamic period across the Mediterranean. In southern Iberia, Argaric societies (about 2200-1550 BC) in today’s Europe were developing complex, often hilltop settlements. They were characterized by planned architecture, advanced metallurgy (especially bronze production), and marked social hierarchies visible in burial practices. These communities often constructed fortified sites, and strategically located settlements controlling surrounding territories.

Increasing evidence suggests that communities in north-west Africa were connected to wider Mediterranean networks. This includes today’s Spain and Portugal. The discoveries at Oued Beht indicate that the Maghreb was not isolated. It participated in broader processes shaping the western Mediterranean.

The defensive system at the site may reflect wider transformations in social organization and regional interaction. There may have been growing competition over resources, control of movement, and more formalized leadership structures.

Rewriting history

The discoveries at Oued Beht highlight how much remains to be learned about the deep history of north Africa. The defensive wall shows that communities here were able to coordinate large construction projects and points to social transformations that are only very tentatively beginning to be recognized.

The medieval remains demonstrate that the ridge was used again many centuries after the prehistoric settlement had been abandoned. Oued Beht was not simply a single-period settlement but a landscape repeatedly selected by different communities across millennia.

As research continues, this location may become key to understanding how societies in the Maghreb evolved from early farming communities to the medieval rural world.

The findings, published in Physical Review Letters, reveal that high-energy streams of particles lose energy as they travel through the material created by these collisions. This effect, called jet quenching, provides an important clue that quark-gluon plasma has formed.

Scientists have previously observed jet quenching in collisions involving much larger atomic nuclei at RHIC, a U.S. Department of Energy (DOE) Office of Science User Facility that operated at Brookhaven National Laboratory from 2000 until 2026. The same effect has also been detected in oxygen collisions at CERN’s Large Hadron Collider (LHC), the European Particle Physics Laboratory.

What makes the new RHIC findings especially significant is that the oxygen collisions occurred at substantially lower energies. The results establish a new lower boundary for the conditions known to produce quark-gluon plasma in this small collision system, helping scientists narrow down what it takes to create matter resembling that of the infant universe.

“The QGP is believed to be the form of matter that filled the early universe — before the protons and neutrons that make up today’s atoms formed. Understanding its properties is essential for uncovering how the early universe evolved in its first microseconds,” said Li Yi, a professor of physics at Shandong University who oversaw the STAR analysis.

“One of the most powerful ways scientists study the QGP is by identifying where its signatures appear — and just as importantly, where these signatures turn off,” she said.

Rongrong Ma, a STAR physicist at Brookhaven Lab, explained why investigating smaller collisions is so valuable.

“This analysis is pushing the boundaries to an extreme case to help us learn something about this jet quenching phenomenon and the conditions needed to generate a QGP.”

How Scientists Recreate Matter From the Early Universe

RHIC was designed specifically to produce and investigate quark-gluon plasma. Traditionally, researchers have accomplished this by accelerating gold nuclei to enormous speeds and crashing them into one another.

Each gold nucleus contains 79 protons and 118 neutrons. When two such nuclei collide with sufficient energy, the boundaries of their protons and neutrons effectively break down, releasing the quarks and gluons inside them. The tremendous energy of the impact also generates thousands of additional quarks and gluons through the conversion of energy into matter, as described by Einstein’s famous equation, E=mc².

The result is a fleeting droplet of primordial matter. It is smaller than an atomic nucleus and survives for less than a billionth of a billionth of a second. Almost immediately, the quarks and gluons recombine into more familiar particles.

Because this exotic substance disappears so quickly, scientists cannot examine it directly. Instead, they use enormous instruments such as the STAR detector to analyze the particles that emerge from the collision.

One particularly revealing clue comes from particles that escape with less energy than expected. Their reduced energy suggests they interacted with the dense mixture of quarks and gluons before escaping.

Researchers pay special attention to jets, which are narrow sprays of extremely energetic particles. These often originate when a quark or gluon from one nucleus strikes a quark or gluon from another, sending two particle jets traveling in opposite directions.

If these paired jets, known as “di-jets,” form near the edge of the collision region, one may have to travel farther through the plasma than the other. Comparing their behavior allows scientists to investigate how the surrounding material drains energy from particles passing through it.

Yi offered a familiar comparison to explain the process.

“You can think of it like shooting a bullet through, say, water,” Yi said. “Because of the resistance, the bullet will slow down. It’s the same thing with these high-energy particles. When they go through the QGP, they will lose energy.”

The Telltale Signs of Quark-Gluon Plasma

For decades, RHIC experiments have provided clear evidence of jet quenching in collisions between gold nuclei. The effect is especially noticeable in jets that travel longer distances through the newly created matter.

Scientists believe these particles lose more energy because they encounter more quarks and gluons before escaping. Jets originating closer to the outer boundary generally pass through less material and experience fewer interactions.

These measurements helped establish that RHIC was successfully producing quark-gluon plasma. Additional evidence came from several other experiments, including an unexpected discovery about how the resulting matter behaves.

Scientists observed that relatively low-energy particles emerging from gold collisions moved collectively, following coordinated patterns. This behavior indicated that the plasma flowed almost like a perfect liquid, with extraordinarily little viscosity or internal resistance.

The discovery was a major surprise when researchers first identified it. Collective flow has since become another important “smoking gun” signature of quark-gluon plasma.

Can Tiny Atomic Collisions Create Primordial Matter?

After establishing these signatures in collisions involving large nuclei, physicists began finding similar collective motion in much smaller systems.

Experiments at the LHC, followed by studies at RHIC, detected collective flow when tiny projectiles such as protons, deuterons, and helium-3 nuclei struck much larger gold or lead nuclei. Researchers at the LHC even observed comparable flow patterns in collisions between individual protons.

These results raised a major question. Did the coordinated particle motion mean that these much smaller collisions were also producing quark-gluon plasma, or could another mechanism explain the observations?

Scientists needed a more definitive sign.

“The next natural question is: Can you see jet quenching, the other key signature that QGP is formed, in these very small collision systems?” asked Brookhaven’s Ma. “We haven’t seen any unambiguous signal, so it’s still kind of under debate in these small systems. We want to search for the smallest system where we can still see an unambiguous jet-quenching signal,” he said.

To investigate, researchers turned to collisions between oxygen nuclei.

An oxygen nucleus contains just eight protons and eight neutrons, making it far smaller than the gold nuclei typically used in RHIC experiments. Oxygen nuclei also have a relatively symmetrical, spherical shape, which makes the geometry of their collisions easier to analyze.

“That means we can more easily look for signs of asymmetric quenching of di-jets produced close to the edge of the circular interaction region in head-on collisions,” said Maowu Nie, an associate professor at Shandong University who contributed to the research.

The scientists could also examine how the outcome changed depending on whether the oxygen nuclei collided directly or merely grazed one another.

“We can also compare how jets behave in head-on central collisions — those that produce the most particles and therefore have the highest likelihood of making a QGP — with collisions where the two oxygen nuclei barely strike each other, which aren’t expected to generate a QGP.”

Nie explained that these grazing collisions are recognizable because they generate far fewer particles than direct impacts.

Oxygen Collisions Reveal a 20% Suppression

To search for jet quenching, the researchers developed an analysis that identifies concentrated streams of high-energy particles and measures how many emerge above a selected energy threshold.

“Our measurements are essentially counting experiments,” said Sijie Zhang, who conducted the research for her master’s thesis under Yi and Nie at Shandong University and is now pursuing a Ph.D. at Yale University.

“We have an algorithm that finds the collimated groups of high-energy particles, the jets, and we count how many of these particles above a certain energy threshold come out.”

Yang He, a former Ph.D. student at Shandong University who developed the analysis framework and now studies oxygen collisions at the LHC, described how the team used different collision types to identify energy loss.

“We compare what we see in back-to-back jets produced near the edge of the system, and we compare what we see in head-on collisions with the counts in non-central collisions where we don’t expect to see quenching.”

The team analyzed STAR measurements from oxygen collisions conducted at RHIC in 2021, at a collision energy of 200 billion electron volts (GeV).

The results revealed a 20% suppression of high-energy particles associated with jet quenching.

Although this suppression was smaller than what physicists typically measure in gold collisions, that difference was expected because oxygen collisions generate much smaller systems.

Crucially, the observed effect was still strong enough to provide clear evidence that quark-gluon plasma had formed.

Comparing Big Bang Matter at Different Energies

Oxygen offers scientists another important advantage. Apart from proton-proton collisions, it is the only nuclear species that has been investigated in collisions at both RHIC and the LHC.

That gives researchers an unusual opportunity to make an “apples-to-apples” comparison between two facilities operating at very different collision energies.

The LHC carried out its first oxygen collisions in 2025. According to results reported by CERN, those experiments revealed both collective flow and jet quenching, providing additional indications that quark-gluon plasma can form in relatively small nuclear collisions.

However, the LHC operated at collision energies exceeding 5,000 GeV, far above the 200 GeV used in the RHIC measurements.

“With a higher energy collision, it’s more likely to create more and more quarks and gluons and a hotter, denser QGP system,” Yi said. “That means the jet quenching signal, if it’s there, should be bigger at LHC compared to RHIC.”

While higher energies make it easier to create the hot, dense conditions needed for quark-gluon plasma, lower-energy experiments may be more useful for determining precisely when those conditions are no longer sufficient.

By reducing the collision energy and using smaller nuclei, physicists can approach the threshold at which the distinctive signatures of primordial matter disappear.

“We are pushing the boundaries in two ways to explore the minimal conditions for creating a QGP, both in terms of the nucleus size by looking at these small oxygen-oxygen collisions, but also by looking at data from lower collision energies — and we still see this crucial QGP signature. That’s a unique feature of our measurement,” Ma said.

New Experiments Could Reveal the Limits of Primordial Matter

Although RHIC has now completed its operations, scientists still have additional oxygen collision data to investigate.

During the collider’s final experimental run, researchers collected measurements using both STAR and sPHENIX, RHIC’s newest detector. The facility concluded its operations with oxygen collisions recorded by both instruments on February 6, 2026.

These remaining datasets could provide further insights into how small a collision system can become, and how low its energy can fall, before quark-gluon plasma can no longer form.

Understanding those limits may help physicists piece together the behavior of matter during the first microseconds of the universe, when quarks and gluons had not yet combined into the protons and neutrons that make up ordinary atoms today.

The research received financial support from the DOE Office of Science, the U.S. National Science Foundation (NSF), and additional international agencies and organizations identified in the scientific publication.

The researchers also used the Open Science Grid, which receives direct support from NSF, along with computing resources at Brookhaven Lab’s Scientific Data and Computing Facilities and the National Energy Research Scientific Computing Center (NERSC), a DOE Office of Science User Facility located at DOE’s Lawrence Berkeley National Laboratory.