A computational approach by St. Jude Children’s Research Hospital scientists promises to make designing T cell-based immunotherapies that target two cancer-related antigens at the same time far easier and faster. Chimeric antigen receptor (CAR) T cells are a type of immunotherapy that reprograms a patient’s immune cells to target a tumor-specific protein antigen. Targeting just one cell surface antigen often is not enough to eradicate the tumor. Thus, scientists have tried to create CARs that target two proteins at once. However, they have encountered problems, including poor CAR expression on the surface of T cells and suboptimal cancer-killing ability.
To address this, St. Jude researchers developed a computational method to screen many theoretical tandem CAR designs and ranked the top candidates for further optimization and validation. The researchers experimentally generated and validated the top-ranked tandem CAR candidates against their chosen targets, demonstrating that the computationally optimized CARs overcame the prior challenges and function better in treating animal models of cancer. The findings were published today in Molecular Therapy.
We have developed and validated a computational tool that can significantly accelerate the design of tandem CAR constructs with improved surface expression and anti-tumor function.”
Giedre Krenciute, PhD, co-corresponding author, St. Jude Department of Bone Marrow Transplantation & Cellular Therapy
While CAR T cells have successfully treated some blood cancers, they have not been as effective in treating solid and brain tumors. One reason is that cancer cells do not uniformly express the same proteins, so CAR T cells targeting a single antigen can miss malignant cells that do not express that protein, leaving them to regrow the tumor and cause a difficult-to-treat relapse. A tandem bi-specific CAR that targets two cancer-related proteins may prevent the original tumor from escaping the treatment, though optimizing their design has been a time-consuming, labor-intensive and expensive challenge in the field.
“Systematic experimental dissection allowed us to first pinpoint the region within the tandem CAR that was problematic for expression and function,” said co-corresponding author M. Madan Babu, PhD, FRS, St. Jude Senior Vice President of Data Science, Chief Data Scientist, Center of Excellence for Data-Driven Discovery director, and Department of Structural Biology member. “This was important and helped guide our efforts as we developed a computational approach for CARs that cleared tumors in our in vivo models more effectively than any single-targeted CAR we tested.”
Clearing tumors with computationally optimized tandem CARs
The computational pipeline predicted a better design for a tandem CAR that targeted the pediatric-brain tumor-related proteins B7-H3 and IL-13Rα2. The original unoptimized version of the bi-specific tandem CAR failed to reach the surface of the T cell, preventing it from contacting its target protein on tumor cells to perform its cancer-killing function. After confirming that the computationally optimized CAR expressed on the T cell surface, the researchers tested it against several single-targeted CARs in mice with tumors that had a mix of cells with both targets, one target or the other, or neither target, mimicking heterogeneous tumors observed in the clinic.
“Our most compelling result is that we completely cleared tumors in four out of five mice with the CAR T cells that had the computationally optimized tandem construct,” said co-first author Michaela Meehl, St. Jude Department of Bone Marrow Transplantation & Cellular Therapy. “By contrast, all heterogeneous tumors treated with single-targeted CAR T cells grew back.”
Additionally, the group showed that they could improve upon the design of several other tandem CARs in the lab. In all cases, the computationally optimized version killed cancer cells better than the non-optimized tandem CARs. The results provide evidence that the design of other bi-specific tandem CARs can benefit from using this computational method to improve and accelerate CAR development efforts.
Creating a generalizable computational tool for CAR construction
“We designed this computational tool to be broadly applicable to many different CARs,” said co-first author Kalyan Immadisetty, St. Jude Department of Bone Marrow Transplantation & Cellular Therapy. “In addition, it can screen roughly 1,000 constructs in a matter of days, greatly speeding up a process that would take many years if researchers created each one in the lab.”
Specifically, to screen so many constructs, the scientists trained an AI-informed algorithm on the structural and biophysical features of known effective CARs. These included predicted properties such as protein folding stability, tendency to aggregate, and other structural and functional features. Together, the program summed these features into a single “fitness” score predicting CAR expression and functionality. CAR designs with the highest fitness score were further optimized to improve protein binding ability.
“Researchers can use our approach to help screen and create better tandem CARs, bringing us closer to the day we can successfully treat challenging tumors, such as pediatric brain cancers,” said Krenciute.
“This work demonstrates the value of creating an intellectual ecosystem that brings together computational and experimental scientists from different disciplines,” said Babu. “Such collaboration drives innovative solutions to major challenges and advances translational applications that serve the St. Jude mission.”
Authors and funding
The study’s other authors are Vikas Trivedi, Pawel Glowacki, Brooke Prinzing, Alejandro Allo Anido, Jorge Ibanez-Vega and Benjamin Leslie, all of St. Jude.
The study was supported by grants from the National Cancer Institute (P30 CA021765 and P30 CA021765), National Institute of Neurological Disorders and Stroke (R01NS121249), the Assisi Foundation of Memphis and ALSAC, the fundraising and awareness organization of St. Jude.
Source:
St. Jude Children’s Research Hospital
Journal reference:
Meehl, M. M., et al. (2025). Computational structural optimization enhances IL13Rα2 – B7-H3 tandem CAR T cells to overcome antigen-heterogeneity-mediated tumor escape. Molecular Therapy. doi.org/10.1016/j.ymthe.2025.07.044.
