Building a safer CAR-T therapy

"We wanted to develop a way to dampen CAR-T cell therapy as a safety mechanism in the event of an adverse reaction in a patient," says Coukos. "To do that we designed CAR-T cells that can be reversibly inactivated with small molecules that can be given systemically and act rapidly."

CAR-T cells are designed to detect specific molecular markers, or antigens, and destroy the cancer cells that bear them. To that end, researchers engineer a chimeric molecule, expressed on a T cell, that is stitched together from the functional units -- or "domains" -- of a few key proteins. The external part of the CAR protein does the antigen detecting. The inner part has two other key components. One is the signaling domain of a protein named CD3-zeta that is absolutely required to activate the T cell. The other is the signaling part of another protein, usually CD28, that supports the proliferation and survival of the activated T cell.

These cellular immunotherapies have been approved for the treatment of some blood cancers, and researchers are working on targeting them at solid tumors. But the treatment has significant risks. CAR-T cells can inadvertently elicit cascading, systemic immune reactions known as cytokine release syndrome, which can cause serious side effects.

Researchers have sought to blunt these risks by, for example, engineering CAR-T cells to commit suicide on demand or require a drug to become activated. "The former approach leads, however, to the waste of a very expensive immunotherapy, while the latter has been challenged by the short half-lives of the drugs," says Irving. "Our approach offers novel and unique solutions to this difficult molecular engineering problem."

To build their "STOP-CAR-T" system, the researchers stuck the CD3-zeta activation domain on one molecule and the antigen-detecting portion on the another. To link the two chains together, so that they'd function as a single unit, they added to each chain the interacting domains of two unrelated proteins that spontaneously pair up inside the cell. The researchers also ensured that the binding could be disrupted by existing small molecules administered systemically. Elegant computational modeling and protein engineering done in Correia's laboratory identified ideal molecular partners for these binding domains and ensured that these newly added binding domains would not interfere with the complex protein interactions within the cell required for the signaling that activates T cells.

The researchers first confirmed in cell cultures that this two-protein CAR-T system -- targeted to a prostate cancer antigen -- worked as well as a similarly targeted but traditionally designed CAR-T system and could be switched off by a drug-like molecule. They then grew tumors expressing that antigen in the flanks of mice and showed that while both types of CAR-T cells could slow tumor growth, only the STOP-CAR-T system's effects could be abrogated with the administration of the small molecule before or after the initiation of CAR-T therapy.

"This really shows that, in principle, we should be able to directly control the activity of the STOP-CAR T cells in patients," says Irving.

The researchers are now developing a STOP-CAR-T system that can be controlled by an approved drug and tweaking the system in various ways to see if they can lower the amount of drug required to control the cells.

"This work itself, and its potential, is really exciting," says Coukos, "but I think it is also illustrative of how well-orchestrated, multidisciplinary collaborations can yield significant scientific breakthroughs. Working with EPFL and our other partners in the region, we hope to bring STOP-CAR-T therapy as quickly as possible to cancer patients."

This study was supported by Ludwig Cancer Research, the Biltema and ISREC Foundations, the European Research Council, the National Center of Competence for Molecular Systems Engineering, The Marie Sklodowska-Curie Actions, Whitaker and the National Research Foundation of Korea.