Enhancing CAR function using chemistry-informed protein design
Leukemia is the most common childhood cancer, and while most patients are cured with standard therapy, those with refractory disease face dire outcomes. The development of a treatment called CAR T cell therapy changed this dramatically; today, more than 80% of children with resistant leukemias achieve complete remission after a single infusion of these engineered immune cells. This is one of the most remarkable advances in the history of cancer.
Despite this success, ~half of children who achieve remission will eventually relapse. In most cases, relapse occurs not because the leukemia has become resistant, but because the CAR T cells themselves stop working - they lose the ability to persist long enough to prevent cancer from returning. Over a decade of research has tried to understand why this happens, and while we have learned a great deal about the molecular programs that drive T cells to fail, no approach has been able to reliably overcome it.
We believe the answer lies in the fundamental design of CAR T cells themselves. The molecular components used to instruct CAR T cells, called signaling domains, were borrowed from the immune system's natural toolkit, where they evolved over millions of years to fight infection. Fighting infection and fighting cancer are fundamentally different challenges, and domains optimized for one cannot reliably achieve the other. To make CAR T cells that can durably cure leukemia, we need to rethink these components from the ground up.
Project Goals
Our goal is to design entirely new signaling components for CAR T cells - ones that are specifically built to instruct durable immunity to cancer. To do this, we recognized a critical but overlooked feature of these signaling components: they lack rigid structure, making them uniquely flexible and redesignable. Their flexibility means their function is governed not by a fixed shape but by their chemical properties. These properties can be systematically studied, mapped and redesigned.
Using an approach we developed for these flexible proteins, we mapped which parts of the CAR signaling domains are essential for long-term T cell function and made the surprising discovery that a region previously thought to be unimportant is actually critical. Our goal is to understand exactly how this region works and to engineer it to improve T cell persistence. In parallel, we have developed a library of new CAR signaling domains built entirely from scratch. We have found that thousands of these entirely man-made domains can instruct T cells to fight leukemia. We will use cutting-edge computational tools and machine learning to identify which chemical features make these synthetic components most effective and iteratively evolve them toward the goal of durable cure.
We expect this work to generate enhanced CAR T cells ready for clinical testing, and to establish a broadly applicable platform for designing better cell therapies for leukemia, brain tumors, and other childhood cancers.
