Scientists identify tumors’ vulnerable point to enhance immunotherapy

Researchers at the USC Viterbi School of Engineering and the Keck School of Medicine of USC have developed a genetic system designed to expose one of the most elusive populations in solid tumors: soft, stem-like cancer cells that can evade immune attack. In a study published in Nature Biomedical Engineering, the team reports that the physical softness of a tumor cell can influence its biology, helping it acquire traits linked to treatment resistance, tumor recurrence and metastasis. By converting that mechanical state into a molecular “kill me” signal, the researchers enabled CAR T cells to recognize and destroy cells that would otherwise be difficult to target.

The work addresses a major limitation of chimeric antigen receptor, or CAR, T-cell therapy. In this treatment, a patient’s T cells are genetically equipped with synthetic receptors that recognize molecules on cancer cells. Once activated, the engineered immune cells attach to their targets and release toxic proteins that damage the cancer cell membrane and trigger cell death. CAR T therapy has produced dramatic results in several blood cancers, but solid tumors remain considerably more difficult to treat because their cells vary widely and are embedded in a complex physical environment that can restrict immune access and promote resistance.

The USC researchers focused on the mechanical properties of tumor cells and their surroundings. Earlier work from the Wang laboratory showed that cancer cells in soft environments can become less susceptible to CAR T-cell killing. The new study links this physical softness to the emergence of cancer stem-like features. These cells can self-renew and generate diverse tumor cell populations, making them important drivers of tumor growth, therapeutic resistance and relapse. Their deformable, springy nature may also make it harder for T cells to establish the firm physical contact required to attack them efficiently.

“Soft environment cancer cells typically have a greater tendency to become stem cell-like cells,” said Peter Yingxiao Wang, chair of the Alfred Mann Department of Biomedical Engineering at USC Viterbi. Unlike stiffer tumor cells, which CAR T cells can more readily engage and mechanically grasp, soft cells may be overlooked during immune surveillance. Lead author Jenny Yunjia Qu compared the problem to trying to pierce a gelatinous material such as Jello or tofu: the target yields rather than providing a stable surface for penetration. This physical mismatch can reduce the efficiency with which T cells form an immune synapse, deliver cytotoxic molecules and kill the cancer cell.

To identify these evasive cells, the team engineered a genetic device called the Mechano-Recorder. The system functions like a biological black box, recording the mechanical conditions experienced by a cell. The researchers found that cancer cells exposed to softer environments displayed substantially elevated intracellular calcium signaling. Calcium ions act as rapid biochemical messengers in many cellular processes, but these signals are normally transient and disappear after the stimulus ends. The Mechano-Recorder captures the calcium response and converts it into a persistent fluorescent signal, creating a stable molecular barcode for cells that have experienced softness.

This conversion allows researchers to identify cells according to a mechanical event that may have occurred hours or days earlier. The recorded signal revealed that softness-associated cells displayed characteristics commonly linked to aggressive disease, including stem-cell-like behavior and markers associated with hypoxia and metastasis. In this way, the tool does more than measure the physical environment in real time: it preserves a record of how that environment has altered cell state. Longwei Liu, an assistant professor of ophthalmology and biomedical engineering at USC and a study co-author, described the approach as taking a molecular snapshot of cancer cells for diagnostic or therapeutic development.

The researchers then rewired the recorder for treatment rather than observation. They replaced its fluorescent reporter with CD19, an antigen already used clinically as a CAR T-cell target in several blood cancers. When a soft, resistant cancer cell generated the relevant calcium signature, the engineered circuit triggered production of CD19 on its surface. The cancer cell was not naturally made more rigid or directly altered to become vulnerable in a physical sense. Instead, its mechanical history was translated into a recognizable molecular beacon. CD19-directed CAR T cells could then identify the reprogrammed cells and attack them selectively.

The strategy was tested in breast cancer cell lines, patient-derived cancer cells and mouse models. According to the researchers, tumors containing the rewired cells showed increased T-cell infiltration and improved killing of the previously resistant, soft cancer stem-like population. The approach was also evaluated in models of glioblastoma, pancreatic cancer and prostate cancer, suggesting that the underlying mechanism may extend beyond breast tumors. Because the system responds to a cellular signaling pathway rather than a cancer type-specific mutation, it could potentially be adapted to recognize other disease-associated states and connect them to therapeutic outputs.

The researchers emphasize that the findings represent a foundational preclinical concept rather than an approved treatment. The work demonstrates that a tumor’s physical properties can be recorded, interpreted and converted into instructions for immune targeting. A mechanical cue such as softness can therefore become an actionable biological signal: cells that once hid within a compliant tumor environment can be marked for destruction by engineered immune cells. The team believes the same design could eventually be adapted to translate other molecular signals into “kill me” or “help me” commands, offering a programmable route for improving immunotherapy against solid tumors.

News Publication Date: 6-Jul-2026

Web References: USC Viterbi School of Engineering; Keck School of Medicine of USC; Wang Lab; https://doi.org/10.1038/s41551-026-01722-7

References: Nature Biomedical Engineering, DOI: 10.1038/s41551-026-01722-7