Chimeric antigen receptor (CAR)-engineered natural killer (NK) cells have shown potent efficacy against hematologic malignancies. In contrast, solid tumors remain difficult to treat due to heterogeneous antigen expression and an immunosuppressive tumor microenvironment that fosters immune evasion and is driven to a large part by checkpoint-mediated inhibition.

To overcome these barriers, we engineered novel dual CAR-NK cells targeting the immune checkpoint PD-L1 and the tumor-associated antigen ErbB2 (HER2). The clinically applicable NK-92 cell line was lentivirally transduced with PD-L1.CAR and ErbB2.CAR constructs. Antitumor activity was assessed across diverse solid tumor models using a comprehensive panel of in vitro and in vivo assays.

Dual CAR-NK cells exhibited potent and selective cytotoxicity against breast, pancreatic, gastric, and lung cancer cell lines expressing either one or both targets. Comparable efficacy was observed in 3D spheroid cultures and against patient-derived primary ovarian cancer cells. Notably, dual CAR-NK cells retained strong cytotoxicity when one antigen was absent or blocked, modeling immune escape through antigen loss - a feature absent in single CAR-NK cells. Mechanistically, each CAR independently activated the MEK/ERK pathway in an antigen-dependent manner, while dual CAR stimulation potentiated the signaling response in a cooperative manner. Furthermore, IFN-γ produced by NK cells upon encounter of resistant cancer cells was sufficient to induce PD-L1 upregulation, thereby enhancing the potency of the dual CAR system. In a PD-L1/ErbB2 double-positive breast cancer xenograft model, dual CAR-NK cells consistently outperformed single-target controls.

Dual targeting of PD-L1 and ErbB2 enhances CAR-NK cell efficacy against refractory solid tumors by providing resilience to antigen heterogeneity and amplifying antitumor signaling through cooperative activation. This approach represents a promising and adaptable platform for clinical translation in solid tumor immunotherapy.

The alternative text for this image may have been generated using AI.