Highlights From AACR IO 2026: Breaching the Tumor Microenvironment Fortress With New CAR T and Chimeras
A tumor is like a strange fortress, hidden in thick fog and surrounded by moats and convoluted ramparts—an environment designed so that the immune cells cannot do their best. This suppressive effect unfortunately impacts the efficacy of chimeric antigen receptor (CAR) T cells and other immune-cell reliant therapeutic approaches. At the AACR Immuno-Oncology Conference (AACR IO), held February 18-21, 2026, multiple studies approached the challenge of breaching the fortress that is the tumor and its suppressive microenvironment from different ingenious angles.
Armored CAR T Cells With a Built-in Safety Mechanism
“CAR T therapy has achieved remarkable success in hematological cancers, but its efficacy in solid tumors remains limited by the immunosuppressive tumor microenvironment, poor T-cell infiltration and persistence, and antigen heterogeneity,” explained Nina Barceló-Genestar, PhD candidate at the August Pi i Sunyer Biomedical Research Institute. When CAR T cells enter a microenvironment that deprives them of supportive growth factors and cytokines, they are pushed towards exhaustion or early death. Barceló-Genestar’s study focuses on one common response to these barriers—arming CAR T cells with additional immune-stimulating functions.
Armored CAR T cells are engineered not only to recognize and kill tumor cells, but also to secrete immune-stimulating molecules that can make the local microenvironment more favorable for an immune attack. Cytokines such as interleukin (IL)-12 or IL-18, for example, can recruit other immune cells, push the immune system toward a more inflammatory state, and help CAR T cells resist suppression. The downside is that these molecules can cause severe systemic toxicity if they spill broadly into the bloodstream, explained Barceló-Genestar. Their strategy was to restrict the release of these molecules to the moments when CAR T cells are engaging a tumor antigen—so that these CAR T cells truly only bring their payloads to bear when encountering a tumor, rather than releasing them everywhere they travel.
The approach that Barceló-Genestar and colleagues came up with leverages a natural regulatory system that already exists inside T cells: microRNAs (miRNAs). miRNA levels are high in resting T cells, but fall when T cells activate. They modified the armored CAR T cells to release their payload only when that drop is detected, ensuring the boost happens only during tumor engagement. “This safety switch approach enables the CAR T cells to autonomously adjust the release of therapeutic molecules according to their activity state,” said Barceló-Genestar.
She and her colleagues identified 58 microRNAs that were abundant at rest but diminish upon activation, then prioritized a handful as the most effective regulators—miR-29c-3p, miR-32-3p, miR-150-5p, and miR-181a-5p. Once they incorporated the corresponding target sites into armored CAR T constructs, the system produced what they were aiming for: strong repression in resting T cells and activation-specific payload expression in vitro and in solid tumor models in vivo. The platform reduced the payload levels in the bloodstream, ranging from 53% to 94% compared with constitutive payload expression, while still improving tumor control and T cell persistence.
This payload could be further fined-tuned, noted Barceló-Genestar. The number and combination of miRNA target sites—the complementary sections of messenger RNA that bind to the miRNA—could be adjusted, with additional sites causing greater repression. This approach worked across multiple payloads (IL‑12, IL‑18, or a secondary CAR), different antigen targets, and in patient‑derived CAR T cells, supporting broad feasibility.
Fine-tuning CAR T Cells Through Cytokine Signaling
Being the frontline defense of the immune system, T cells can differentiate into many states, some optimized for immediate cytotoxicity and others optimized for longevity and memory. Those states are influenced by cytokine signaling, and as mentioned in the previous section, cytokine signaling is often distorted inside solid tumors, where supportive cytokines can be scarce. As a result, CAR T cells frequently lose function or fail to survive long enough to mediate durable tumor control. “We wanted to engineer cytokine-like signaling inside CAR T cells in a programmable way, so persistence and function are no longer entirely at the mercy of the tumor microenvironment,” said Wan Sang Cho, PhD, postdoctoral researcher at Stanford University.
To do this, Cho and his colleagues looked at the problem from a different angle—instead of approaching cytokine signaling as a fixed menu of natural pathways, they considered the building blocks that go into producing the result of each pathway. Natural cytokine receptors achieve diversity through combinations of intracellular motifs—short sequences of amino acids that exist on the tail of signaling proteins—to recruit different signaling proteins. The team’s hypothesis was that recombining these motifs in different ways could generate controllable signaling outputs, and thus controllable CAR T-cell phenotypes.
They built a platform of constitutively active synthetic cytokine receptors (SCRs) that provided continuous signaling without needing the cytokine to be added. They then recombined 14 signaling motifs to create a large combinatorial library of roughly 30,000 SCRs and experimentally tested about 450 in CAR T cells, measuring memory, cytotoxicity, and proliferation.
They found that CAR T cell traits do not neatly fall into these boxes, instead, they range from a spectrum with different features at each end. Across their screened designs, they were able to quantify an immunological relationship: a tradeoff between memory and cytotoxicity. In their system, STAT5-heavy signaling tended to drive effector-like cytotoxicity, while more balanced STAT1/STAT3 signaling supported immunological memory alongside killing.
“By integrating experimental data with neural network models, we identified features that promote memory, cytotoxicity, and autonomous proliferation across the full SCR design space,” explained Cho. In addition to identifying combinations that would result in desirable features in CAR T cells, Cho’s team also found ways to suppress features such as unsafe autonomous CAR T-cell proliferation by reducing strong STAT5 and Shc signaling combinations.
“These findings highlight the potential of engineering cytokine signaling to fine tune immune cell functions and phenotypes, adding an important new layer of programmability to immune cell engineering for therapeutic applications,” said Cho.
Exploiting the Sugar Coating on Cancer Cells
“Glycans are sugar structures that coat the surface of all our cells and play essential roles in normal biology,” said Megan Priestley, PhD, postdoctoral researcher at the Massachusetts Institute of Technology. In cancer, however, these structures are exploited to help tumors evade the immune system, she explained. Cancer cells display abnormal glycans that engage immunosuppressive glycan-binding receptors, called lectins, on immune cells, forming glyco-immune checkpoints, a topic that was discussed a length at the inaugural AACR IO in 2025. These checkpoints can contribute to immune evasion and therapeutic resistance—limiting the effectiveness of therapies that rely on immune engagement like CAR T cells.
While glyco-immune checkpoints are a prominent mechanism of immune evasion and resistance, therapeutically targeting these interactions has been challenging, said Priestley. One reason is that glycan-lectin interactions can be relatively low affinity, and the exact identity of relevant glycan ligands can be complex. Another is that systemic blockade of immune checkpoints, even protein-based ones, can produce immune-related adverse events, so targeted approaches can offer a better safety profile. In this study, Priestley and her team introduce a new molecular architecture called antibody-lectin chimeras (AbLecs), designed to solve these constraints by localizing glycan blockade where it matters most: at the tumor surface and at the immune synapse where immune decisions are made.
“AbLecs are antibody-like molecules with two key parts: an antibody fragment, which is an antigen-binding (Fab) domain that guides the drug to cancer cells by binding tumor antigens, such as HER2, and a lectin domain that binds cancer-associated glycans and blocks their interaction with inhibitory lectin receptors on immune cells,” explained Priestley. Once bound to the cancer cell, the AbLec’s Fc domain engages Fc receptors on immune cells, initiating an antitumor immune response.
Priestley and colleagues found that AbLecs promote immune-mediated cancer cell killing both in vitro and in vivo, and were able to outperform conventional monoclonal antibodies targeting the same antigens. They built AbLecs by combining trastuzumab (Herceptin)—a monoclonal antibody used to treat breast cancer—with lectin domains from Siglec-7 and Siglec-9—immune inhibitory proteins associated with cancer aggressiveness across multiple cancer types. These AbLecs showed nanomolar-level binding efficacy and elicited stronger immune‑mediated antitumor responses than trastuzumab alone or combination immunotherapy that consisted of trastuzumab and Siglec “decoy receptors,” which were antibody constructs that only have the lectin-binding domain and lack the antigen-binding domain.
“This approach can be readily adapted to treat numerous types of cancer via different mechanisms of action and so have the potential to broaden the reach of cancer immunotherapy, particularly to patients who are ineligible for, or unresponsive to current treatments,” concluded Priestley.
