The new computational platform can design synthetic protein receptors from scratch

Cancer immunotherapy, especially using T-lymphocytes, shows promise in the treatment of blood cancers. Engineered T cells, especially those equipped with chimeric antigen receptors (CAR-T cells), have revolutionized cancer treatment. But while they have shown impressive results against some blood cancers, they have struggled to affect solid tumors such as those in the breast, lung and prostate.

The tumor microenvironment is the problem

An important problem is the tumor microenvironment (TME), which is a mixture of cells and molecules that can dampen immune responses. In most solid tumors, inhibitory signals predominate, while helper signals (which tell T cells to keep going) are weak or completely absent. Since modified T cells rely on these environmental cues to remain active and functional, they often fall short. This has led scientists to explore ways to create extra receptors on T cells so they can receive tumor-specific signals and respond with added force.

Researchers have tried to create receptors that can sense and react to the TME, but designing them has been difficult because making custom signaling proteins is a complex endeavor. Meanwhile, most current methods to do this rely heavily on trial and error, which ultimately makes it difficult to control how these synthetic receptors will ultimately behave when deployed against a tumor.

Computational design solution

Now, a team led by Patrick Barth at EPFL and Caroline Arber at UNIL-CHUV has developed a computational platform for designing synthetic protein receptors from scratch. These receptors, called T-SenSERs (tumor microenvironment-sensing switch receptors), are designed to detect soluble signals found in tumors and convert them into costimulatory or cytokine-like signals that enhance T cell activity. When combined with conventional CAR-T cells, the synthetic receptors enhanced their antitumor effects in lung cancer and multiple myeloma models.

The study is published in Nature Biomedical Engineering.

The computational platform can assemble artificial receptors by designing and combining different protein domains, like building with molecular Legos. Each receptor includes an outer region that binds a tumor-associated signal, a middle region that transmits that signal across the cell membrane, and an inner region that activates useful functions within the T cell.

What differentiates this approach from current protein design approaches is that it does not treat proteins as rigid structures. Instead, it models them as dynamic, shape-changing machines, allowing researchers to see, for the first time, how signals travel through these synthetic receptors to control cell behavior.”

Patrick Barth at EPFL

Testing the T-SenSERs

Using the platform, the team created two T-SenSER families: one that responds to VEGF, a protein that promotes blood vessel growth and is common in tumors, and another that responds to CSF1, a protein that negatively affects the behavior of immune cells in tumors. They designed 18 versions and selected the best performers based on simulations and laboratory tests.

When tested, T cells equipped with both CAR and T-SenSER responded more strongly to tumors than CAR-T cells alone and showed ligand-specific activities that closely mirrored the signaling programs encoded by the design method.

The VEGF-sensing version (called VMR) activated the T cell only when VEGF was present, while the CSF1-sensing version (CMR) provided a small baseline boost even without CSF1, but increased its effect in the presence of the ligand. In mouse models of lung cancer and myeloma, T cells with these synthetic receptors showed improved tumor control and longer survival.

Importantly, the team found that their design method allowed them to tailor the receptor’s behavior, choosing whether it should always be activated, ligand-dependent, or somewhere in between.

“This study represents the first demonstration of the computational design of single-pass, multidomain receptors with programmable signaling functions and paves the way for the accelerated development of synthetic biosensors with tailored sensing and response capabilities for basic and translational cell engineering applications,” says Barth.