Since the FDA approved the first chimeric antigen receptor (CAR) T-cell-based therapies for advanced lymphoma and acute lymphoblastic leukemia in 2017, the success of these therapies has been limited primarily to hematologic malignancies.1–3 Despite consistent efforts in the field, these successes have not been replicated for solid tumor malignancies. Key limitations include identifying tumor-specific cell-surface proteins (i.e., tumor-associated antigens [TAA]), as well as challenges facing antigen escape and accessing the tumor microenvironment.3 Ideal targets should be specific to tumor cells, highly expressed, and also absent in healthy tissue. These constraints have rendered many solid cancers to be considered unapproachable with the current armamentarium of CAR-T therapies.
Two paradigm-shifting reports were published in 2023 — one by Mark Yarmarkovich, PhD, and colleagues at Children’s Hospital of Philadelphia4 and another by Rosa Vincent, PhD, and colleagues at Columbia University in New York.5 Both studies highlight the potential of novel engineering approaches that allow CAR-Ts to target and kill solid tumors by targeting intracellular proteins (Yarmarkovich and colleagues) and leveraging the ability of bacteria to selectively colonize and “paint” difficult-to-reach solid tumors (Vincent and colleagues).
An important difference between CAR T-cell and cellular therapies that rely on manipulation of a native T-cell receptor (TCR) is that CAR T cells can only target proteins naturally expressed on the cell surface, while TCR-engineered T cells are designed to be stimulated by intracellular peptide fragments presented in the context of major histocompatibility (MHC) molecules, which are encoded by specific human leukocyte antigen (HLA) genes.6 Receptors on these T-cell types are generated in fundamentally different ways: CAR T cells use antibody derivatives (single-chain variable fragments [scFVs]) as their weapon, whereas native T cells (or TCR-engineered T cells) use naturally evolved receptors directed against peptide-MHC (pMHC) complexes.
As most driver proteins responsible for malignant cell growth live and do their work intracellularly, a significant number of potential tumor-related antigens have been unavailable for traditional CAR-T engineering approaches. Dr. Yarmarkovich and colleagues have upended this paradigm by constructing a specific scFV that can target peptides derived from an intracellular protein (PHOX2B) specifically expressed by neuroblastomas. To achieve this, they used a protein display platform7 containing a library of more than 1011 scFVs, followed by computational and experimental selection to minimize cross-reactivity with other peptide-MHC complexes. Furthermore, the peptide-centric (PC) CAR T cells generated using this scFV construct resulted in selective killing of neuroblastoma cells in vitro as well as eradication of tumors in mice. What is particularly remarkable is that naturally occurring T cells capable of targeting PHOX2B pMHC were not identified (presumably because PHOX2B is an unmutated self-protein, meaning that reactive T cells would undergo negative selection in the thymus). Similar to TCRs, these scFV receptors would also be restricted to specific HLA haplotypes, though there may be enough similarity in various haplotypes for one receptor to target multiple MHCs and cover a significant proportion of the population.4 Challenges remain in proving that there is indeed minimal cross-reactivity with other pMHCs, as these scFVs were not naturally evolved.8 Nonetheless, we anticipate that this approach will gain increasing traction as CAR-T engineering continues to improve.
Given the challenge in identifying TAAs on solid tumors, the Columbia team took a different approach, tagging the tumor cells with a synthetic antigen that could then serve as a target for CAR T cells.5 To accomplish this, they leveraged the ability of engineered bacteria that preferentially grow in the hypoxic necrotic tumor microenvironment9 to deliver synthetic CAR targets directly into the tumor core. While the concept of using engineered bacteria as intratumoral bioreactors to deliver active compounds is not new,10, 11 this is the first study to demonstrate the potential for probiotic-guided CAR T cells (ProCARs) engineered to respond to antigens delivered by tumor-colonizing bacteria. The authors achieved this by designing a synthetic target (Tag), which consisted of a green fluorescent protein derivative known to mediate CAR-T responses fused with the heparin-binding domain of placenta growth factor (PIGF-2), which enables anchoring to extracellular matrix components found in high abundance in tumor stroma. Mice bearing solid tumors that were injected with bacteria producing GFP derivative alone did not respond to ProCAR infusion, while those exposed to a strain producing Tag demonstrated tumor response and improved survival. These benefits were further augmented by co-expressing an activated mutant of CXCL16, a chemokine that recruits T cells to local tissue. While these findings are intriguing, the mode of bacterial delivery remains to be optimized: Systemic injection of Tag strains into mice was assessed as potentially feasible, though will likely require further engineering to attenuate potential inflammatory responses before being deemed viable in the clinical setting.
The findings of Dr. Yarmarkovich and colleagues extend the range of targetable tumor antigens to include intracellular peptides by integrating a robust protein display platform with computational and experimental optimization strategies that screen out off-target peptide-MHC complexes. Those of Dr. Vincent and colleagues demonstrate the ability to create synthetic targets by leveraging engineered bacteria with an affinity to hypoxic tumor microenvironments to deliver a targetable payload that “paints” the tumor cell surface. Together, these studies demonstrate the ability for creative engineering techniques to redefine the therapeutic potential of CAR T cells for otherwise difficult tumor targets.
Drs. Yeh and Markey indicated no relevant conflicts of interest.