When baking cookies, sometimes a kitchen mishap will result in failure. Often, the answer is obvious — too much flour, not enough sugar, or forgetting to preheat the oven. Other times, finding the answer requires reevaluating every step, searching for what was missed. This is especially true when something unexpected happens, such as a two-hour baking time for what normally takes 15 minutes. The answer may be more complicated, with extenuating circumstances keeping the oven from behaving as expected.
Chimeric antigen receptor (CAR) T cells, a type of immunotherapy that reprograms a patient’s own cells to kill cancer cells, have been used to cure relapsed blood cancers, such as leukemia. Due to these early successes and experiments in animal models, the expectation was that CAR T–cell therapy would translate to other difficult-to-treat tumors, such as high-grade gliomas in the brain. However, these approaches were much less effective than predicted, akin to the mysteriously ineffective oven.
When baking cookies, one of the most important factors in their success is easy to miss: the air that surrounds them. This is similar to how certain immune cells are easy to overlook, though they exist around the tumor, making up the tumor immune microenvironment (TIME). Air surrounding cookies must be of a certain density, so baking at an elevation where the air is thinner increases the time needed in the oven. There may be nothing wrong with the oven — or CAR T cells — the local environment may be different than expected.
St. Jude researchers discovered an underappreciated role of TIME on the efficacy of high-grade glioma-targeted CAR T cells. The work was published recently in Cancer Research Communications.
“I wanted to understand, when we inject the CAR T cells, what happens to the other immune cells,” said corresponding author Giedre Krenciute, PhD, St. Jude Department of Bone Marrow Transplantation and Cellular Therapy. “Do they help CAR T cells, or do they inhibit anti-cancer activity?”
The researchers found that the key appears to be immune cells in the TIME, particularly macrophages. When they depleted macrophages in an immunocompetent mouse model, the anti-cancer activity of CAR T cells failed to control the tumors.
“We saw that the most effective CAR T cells recruited a specific type of macrophages,” Krenciute explained. “And the most interesting thing we found about those specific macrophages is that you cannot classify them into traditional groups.”
Macrophages are typically classified as either M1 or M2 — groups based on functional and transcriptional profiles. However, the St. Jude scientists performed single-cell RNA sequencing that revealed a novel group that did not fit either the M1 or M2 criteria yet appeared responsible for affecting CAR T efficacy. They were only able to find that population after a long and arduous process to create the right model system.
A true novelty of the study was its use of immunocompetent mice that mimic human disease and resistance factors. Traditionally, most CAR T–cell work is performed with human cell lines or in immunodeficient mice, targeting human proteins. To truly investigate the impacts of the immune system, the researchers created an immunocompetent model targeting murine versions of proteins, using murine CAR T cells, to be as informative as possible.
“The really big novel and new part of what we’re doing is that we are more closely replicating patient’s cancers,” said first author Dalia Haydar, PharmD, PhD, formerly a postdoc at St. Jude and now faculty at Children’s National Hospital. “A patient will have an immune system, so using this system may enable us to better predict how those CAR T cells will function in patients.”
Armed with the immunocompetent model, the St. Jude group could compare the activity of CAR T cells, both in vitro and in vivo, in models of high-grade glioma to find where the TIME had the most influence. They compared six different CAR molecules to each other, which all had different structural domains (like testing different fillings in cookies) to see which worked best. Each had shown promise during in vitro experiments, as well as in vivo using immunodeficient mice, but never in the context of a mouse with a full immune system.
“The results were completely different, even though they all worked really well in vitro,” Krenciute said. “Those signaling domains that we changed somehow affected the TIME and, as such, dictated the therapeutic efficacy.”
As an example, two of the CAR constructs showed alternate promises. One, which has a 4-1BB ligand (L) costimulatory domain in trans (not part of the CAR molecule), performed best in vitro. Another, which has a CD28 costimulatory domain, had a modest performance. However, during the in vivo experiment, CD28-based CAR was the only construct to truly contain tumor growth, while the CAR that included a 4-1BBL showed the least control.
“We showed it’s not enough to generate CARs and just compare them in a cell culture setting,” Haydar said. “You need to evaluate the efficacy of those different designs in a real, immunocompetent, tumor-bearing mouse to know how those CARs will function.”
The researchers showed that the TIME impacts CAR T–cell therapy, guided by a unique population of macrophages. With this study, the St. Jude group is ushering in a different way of evaluating T cells, accounting for that missing component.
“This is just the beginning,” Krenciute said. “I think there is going to be an influx of papers on how the TIME matters, and we should all study its effects to improve CAR T–therapy.”
Like trying to bake cookies at a high elevation and needing to adapt the oven time, adapting to looking at the TIME in immunocompetent models will be a process. But the potential promise of that adaptation could have far more wide-reaching consequences than a tray of perfectly baked cookies.
“I believe we have the perfect system to understand how CAR T cells are working,” Haydar said. “While this study was done in one model, our tools will facilitate understanding the effects of the TIME on other pediatric brain tumors. Eventually, we’ll be able to use it to decide how to make the best working CAR that could potentially cure brain tumors.”