The device, called a Bacteria-Releasing Mucoadhesive Sheet, or BRMS, was developed by Jihun Lee, Sana Ashraf, and colleagues working with Sukho Park at institutions including the Korea Institute of Science and Technology and Daegu Gyeongbuk Institute of Science and Technology. Its design responds to three formidable barriers that defeat conventional oral drug delivery in the gut: the acidic environment of the stomach, the sticky mucus layer that coats the intestinal wall, and the tight junctions that seal the epithelial lining. Rather than swallowing a pill and hoping it survives the journey, clinicians would deploy the BRMS directly onto a tumor using an endoscope or a magnetically steered capsule robot, keeping the sheet sealed and dry until the moment of release.

The sheet’s architecture is a study in layered functional engineering. An outer unrolling layer made of poly(ethylene glycol) dimethacrylate swells dramatically when it contacts intestinal fluid, generating a bending moment that causes the rolled sheet to unfurl within roughly 100 seconds and drape itself conformally over curved mucosal surfaces. Beneath it sits a guard layer of tri(ethylene glycol) dimethacrylate, a nearly impermeable polymer that blocks bacteria from escaping into the gut lumen and enforces strictly one-way release toward the tumor. The therapeutic layer itself is split into two zones: a peripheral ring of alginate and skim milk that provides powerful mucoadhesion, and a central reservoir containing freeze-dried, genetically engineered Salmonella together with the sugar L-arabinose.

The mucoadhesion is not a minor detail — it is the linchpin of the entire strategy. In the turbulent environment of the intestine, where peristaltic waves and continuous fluid flow scour the mucosal surface, free bacteria or drug particles are swept away within minutes. Quantitative tests on porcine small intestinal tissue showed that after two minutes of contact, the BRMS required a detachment force of 0.544 newtons, roughly four times greater than a gelatin control sheet commonly used in drug delivery research. The alginate-rich peripheral region achieves this grip through abundant carboxyl and hydroxyl groups that form dense hydrogen-bonding networks with mucin, the glycoprotein that gives mucus its adhesive character. Skim milk, meanwhile, serves as a cryoprotectant that shields bacterial cell membranes from ice-crystal damage during freeze-drying, allowing the bacteria to be stored dry and revived on demand.

The therapeutic payload is as sophisticated as the carrier. The researchers used an attenuated Salmonella Typhimurium strain, engineered with deletions in the aroA, aroD, rcsB, and asd genes to reduce virulence, and equipped it with a plasmid that fuses the anticancer peptide VC1 to FlgM, a protein naturally exported through the flagellar type III secretion system. VC1, also known as α-conotoxin Vc1.1, is a 16-amino-acid peptide originally discovered in the venom of the marine cone snail Conus victoriae. It blocks the α9α10 nicotinic acetylcholine receptor, a receptor implicated in cancer cell proliferation, survival, and migration. When the sheet hydrates at the target site, the co-encapsulated L-arabinose switches on the bacterial secretion machinery, and the Salmonella begin pumping out VC1 into the tumor’s extracellular space.

What sets this system apart from passive drug formulations is what happens next. Conventional nanoparticles and small-molecule drugs rely on diffusion, a process that stalls in the dense extracellular matrix and elevated interstitial fluid pressure of solid tumors. In confocal microscopy experiments on three-dimensional CT-26 colon cancer spheroids roughly 400 micrometers across, a doxorubicin-loaded control sheet delivered its payload only to the outer rim of the tumor model, with signal fading to nearly nothing beyond about 21 percent of the spheroid’s radius. The Salmonella released from the BRMS, by contrast, distributed throughout the entire spheroid, propelled by their flagella and guided by their natural preference for the hypoxic, immunosuppressed conditions found in tumor cores. In a Transwell assay simulating mucus barriers and fluid wash-out, the sheet achieved bacterial penetration roughly 130-fold higher than a free bacterial suspension of the same dose.

The cytotoxicity results under physiologically mimetic conditions were equally striking. When CT-26 colon cancer cells were separated from treatments by a motility agar barrier and subjected to a wash-out step, neither the doxorubicin-loaded sheet nor the free Salmonella suspension produced meaningful cell death — both were defeated by the physical barrier and fluid flow. The fully functional BRMS carrying VC1-induced Salmonella, however, killed the largest fraction of cancer cells, combining the bacteria’s intrinsic oncolytic activity and nutrient competition with the anticancer action of locally secreted VC1. The effect extended across species: supernatants from VC1-secreting bacteria significantly reduced viability not only in mouse CT-26 cells but also in human SW480 and HT-29 colorectal cancer lines.

Perhaps the most clinically compelling demonstrations came from the delivery experiments. Using a standard commercial colonoscope fitted with a soft Ecoflex cap that stays sealed during navigation and opens only when forceps push the sheet out, the team deployed BRMS units onto porcine intestinal tissue mounted in a 3D-printed phantom. The sheets released, self-unrolled, and adhered conformally to the mucosa within view of the endoscopic camera, and a separate test inside an intact, uncut porcine intestinal lumen confirmed the approach works under realistic anatomical conditions. The researchers also loaded four rolled sheets into a custom magnetically actuated capsule robot, steered it wirelessly using a six-coil electromagnetic actuation system, and sequentially deposited all four units at spatially distinct target sites — a capability that could allow multiple lesions to be treated in a single procedure.

In vivo, the platform delivered its most dramatic result. In BALB/c mice bearing subcutaneous CT-26 tumors, mice implanted with the fully activated BRMS — bacteria plus L-arabinose induction — showed progressive tumor regression rather than mere growth delay, ending the two-week study with markedly smaller tumors than every control group. Crucially, mice that received the same dose of VC1-secreting Salmonella as a free suspension showed tumor growth indistinguishable from untreated controls, a direct demonstration that the therapeutic benefit came from the sheet’s ability to retain bacteria at the lesion, not from the bacteria alone. Histology revealed extensive necrosis and elevated apoptotic cell populations in treated tumors, and no significant body weight loss or systemic toxicity appeared in any group.

The authors are candid about the hurdles that remain before the BRMS reaches patients. The L-arabinose inducer loaded within the sheet proved insufficient for autonomous high-level VC1 secretion in vitro, forcing daily intraperitoneal injections in the mouse study — an approach that would undermine the convenience of localized therapy in the clinic. Future iterations may engineer bacteria that secrete VC1 constitutively or respond to tumor-specific cues such as hypoxia or acidity, or deliver the inducer orally. The in vivo model also placed tumors under the skin rather than in the gut, so orthotopic studies in large animals will be needed to validate unrolling, adhesion, and bacterial kinetics in a real gastrointestinal environment. Biosafety, too, demands attention: although the attenuated strain caused no observable adverse effects, infection risk in immunocompromised patients may ultimately favor swapping Salmonella for clinically validated probiotic strains such as engineered E. coli Nissle 1917. Even so, the BRMS stands as a vivid example of what happens when synthetic biology, materials science, and surgical robotics converge — a living drug factory, glued to a tumor, swimming medicine into places no molecule could reach on its own.

Subject of Research: A mucoadhesive bacteria-releasing sheet for localized, penetrative therapy of gastrointestinal cancer using engineered Salmonella

Article Title: A bacteria-releasing mucoadhesive sheet for localized delivery and penetrative therapy of gastrointestinal cancer

Article References: Lee, J., Ashraf, S., Kim, E., Lee, H.-J., Park, J., Jeon, H. J., Suh, S., & Park, S. (2026). A bacteria-releasing mucoadhesive sheet for localized delivery and penetrative therapy of gastrointestinal cancer. Materials Today Bio, 41, Article 103710. https://doi.org/10.1016/j.mtbio.2026.103710

Image Credits: AI Generated

DOI: 10.1016/j.mtbio.2026.103710

Keywords: gastrointestinal cancer, Salmonella, mucoadhesive sheet, drug delivery, bacterial therapy, VC1 conotoxin, capsule robot, endoscopy, tumor penetration, biotechnology, oncology, hydrogel

CAR therapy works by equipping an immune cell with a synthetic receptor that recognizes a chosen molecule on the surface of a cancer cell. The receptor typically contains an antibody-derived binding region, known as a single-chain variable fragment, or scFv, connected to signaling components that activate the engineered cell. In CAR-T-cell therapy, this design enables T cells to identify and kill malignant cells. CAR macrophages use the same broad principle but assign the task to macrophages, immune cells naturally adapted to engulf cellular material through phagocytosis. Once activated, a macrophage can surround a target, internalize it and digest it in an intracellular compartment. Yet solid tumors present a difficult environment for these cells. They may be poorly recruited into the tumor mass, become functionally suppressed after arrival or fail to recognize cancer cells that express only small amounts of the target antigen. A therapy dependent on one receptor-antigen interaction can therefore lose effectiveness when tumors alter or reduce that molecular marker.

The researchers’ solution was to combine the target recognition and engulfment machinery of CAR macrophages with the mobility and destructive chemistry of neutrophils. Neutrophils are among the first immune cells recruited to sites of infection or tissue damage. They respond rapidly to chemical gradients, migrating toward signals released by inflamed or injured tissue. They can also deploy neutrophil extracellular traps, or NETs: web-like structures composed largely of DNA and associated proteins that immobilize threats outside the cell. During this process, and through other antimicrobial responses, neutrophils release reactive oxygen species. These chemically reactive molecules can damage biological structures, although their effects must normally be tightly controlled to avoid harming healthy tissue. By fusing neutrophils with CAR macrophages, the investigators sought to create a single cell-like therapeutic system that could combine directed tumor recognition, active phagocytosis, chemokine-guided movement and neutrophil-derived attack mechanisms. The resulting S-CAR-Ms were not simply macrophages carrying a second drug; they were designed as an integrated cellular platform whose functions could reinforce one another.

A central finding was that the fused cells accumulated in tumors more efficiently than conventional CAR macrophages. The reported explanation is chemokine-driven migration, a process in which immune cells detect concentration gradients formed by signaling proteins released from tumors and surrounding tissues. Neutrophils are highly responsive to these gradients, and their migratory behavior appears to remain functionally important after integration with CAR macrophages. This matters because the number of engineered cells administered to a patient is not the same as the number that reaches the malignant tissue. Solid tumors can contain dense extracellular matrix, abnormal blood vessels, regions of low oxygen and immunosuppressive cells that collectively form a physical and biochemical barrier. A cell that recognizes cancer in a laboratory dish may therefore perform poorly if it cannot reach the tumor or move through it. The enhanced accumulation of S-CAR-Ms suggests that neutrophil properties may help overcome one of the earliest bottlenecks in cell therapy: getting therapeutic cells out of the circulation and into the disease site in sufficient numbers.

Once inside tumors, S-CAR-Ms appeared to exploit a second advantage inherited from their neutrophil component. The release of NETs and reactive oxygen species increased the exposure of phosphatidylserine, or PtdSer, on tumor cells. PtdSer is a phospholipid normally concentrated on the inner surface of the plasma membrane. When a cell is stressed, damaged or undergoing programmed cell death, PtdSer can become exposed on the outer membrane, where it acts as an “eat me” signal for phagocytic cells. Macrophages recognize this signal through receptors and bridging molecules, including the MerTK pathway. The investigators found that the neutrophil-derived activity made more tumor-cell material visibly available to this clearance system. In practical terms, the S-CAR-Ms could identify cancer through their engineered scFv receptor when the selected antigen was present, but they could also recognize and engulf damaged tumor material through the PtdSer–MerTK route. That dual recognition system is important because it reduces dependence on a single molecular label.

The two pathways may also create a self-reinforcing cycle inside the tumor. CAR recognition can bring the engineered macrophage into close contact with an antigen-bearing cancer cell, while neutrophil-derived reactive molecules can injure nearby tumor cells and expose PtdSer. The macrophage can then engulf cellular debris through its natural clearance machinery, even when that debris contains little of the original CAR target. This distinction addresses a major problem in cancer immunotherapy known as antigen escape. Tumors are genetically diverse populations rather than uniform masses. If treatment eliminates cells with abundant target antigen, pre-existing or newly selected variants with low levels of that antigen may survive and repopulate the tumor. A therapy that combines antigen-specific recognition with a broader damage-associated signal could continue to remove cells that would otherwise slip past the CAR receptor. The study reports that S-CAR-M treatment triggered antigen spreading, meaning that the immune response expanded from the original targeted antigen to additional tumor-associated targets released or revealed as cancer cells were destroyed. This process could make the attack less vulnerable to the tumor’s molecular evolution.

The researchers tested the approach in both syngeneic and xenograft mouse models, two experimental systems that answer different questions. Syngeneic models use tumor cells and immune cells from genetically compatible animals, allowing investigators to study treatment in the presence of an intact immune system. Xenograft models implant human or otherwise foreign tumor cells into mice, often in settings designed to permit tumor growth despite immune incompatibility. Across these models, the study reports that a single dose of S-CAR-Ms reduced tumor burden, limited the spread of cancer to distant sites and helped prevent recurrence. Those findings are particularly notable because recurrent disease and metastasis are responsible for much of cancer’s lethality, while many experimental treatments show their strongest effects only against established primary tumors. However, the mouse results cannot yet establish whether the cells will behave similarly in people. Human tumors vary widely in their chemokine signals, antigen expression and tissue architecture, and immune-cell fusion products must also be manufactured consistently and tested for safety.

The design nevertheless highlights why solid tumors have remained a difficult frontier for engineered-cell therapy. In blood cancers, therapeutic cells can circulate through a relatively accessible compartment and encounter malignant cells directly. Solid tumors are more like hostile ecosystems, with abnormal vasculature, low nutrient and oxygen levels, high interstitial pressure and suppressive signals that can blunt immune function. Macrophages are naturally abundant in many tumors, but tumor-associated macrophages are often reprogrammed into states that support cancer growth, tissue remodeling or immune suppression. Engineering them with a CAR can redirect their recognition, but it does not automatically solve problems of trafficking or local activity. Neutrophils offer a different set of biological tools, yet their inflammatory molecules can also cause collateral tissue injury if unleashed without adequate control. The therapeutic promise of S-CAR-Ms therefore depends not only on their ability to kill tumor cells, but also on whether their activity can remain localized, whether they persist for an appropriate period and whether their manufacture avoids unwanted activation or inconsistent cell states.

The next steps will require detailed safety and translational studies before the approach can be considered for patients. Scientists will need to determine how the fused cells are produced, how stable the fusion state remains, how long the cells survive after infusion and whether they can be controlled or eliminated if severe inflammation develops. The balance between tumor-damaging reactive oxygen species and injury to healthy tissue will be especially important. Researchers will also need to test whether the method works across different cancer types and target antigens, and whether the chemokine signals that attract neutrophil-integrated cells are present in human tumors at useful levels. Even with those questions unresolved, the study offers a striking example of immune engineering moving beyond the idea of giving one cell one receptor and one job. By combining receptor-guided recognition, neutrophil-like migration and oxidative damage with macrophage-mediated clearance and antigen spreading, S-CAR-Ms are intended to confront cancer as a moving, heterogeneous target. In mice, that integrated strategy produced a broader response than conventional CAR macrophages; whether it can translate into a safe treatment for solid tumors will determine the significance of the advance.

Subject of Research: Neutrophil-integrated syncytial chimeric antigen receptor macrophages for cancer immunotherapy

Article Title: Neutrophil-integrated syncytial CAR macrophage for cancer immunotherapy

Article References: Tian, T., Zhao, S., Tian, T. et al. “Neutrophil-integrated syncytial CAR macrophage for cancer immunotherapy.” Nature Immunology (2026). https://doi.org/10.1038/s41590-026-02615-2

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41590-026-02615-2

Keywords: cancer immunotherapy, CAR macrophages, neutrophils, solid tumors, phagocytosis, antigen escape, antigen spreading, neutrophil extracellular traps