NK cells are the most common and successful non-T-cell source for CAR therapy to date. NK cells, like T-cells, exert potent cytotoxicity against target cells, making them an attractive candidate for CAR modification. Unlike T-cells, they do not have a TCR and do not exert significant GVHD, making them an attractive allogeneic product. NK cells recognize and kill their targets through a variety of mechanisms, including through activating receptors such as NKG2D, NKp44, NKp46, NKp30, and others, which can be triggered by a variety of ligands, many of which are upregulated by cellular stress and are thus often naturally found on tumor cells. Additionally, NK cells express the Fc receptor FcγRIIIa (CD16) and can recognize and kill cells coated with antibodies through antibody-dependent cellular cytotoxicity (ADCC). Thus, their ability to kill through multiple mechanisms in addition to the CAR may reduce the risk of tumor antigen escape. NK cells also possess inhibitory receptors, such as those that prevent killing of MHC-expressing self-cells, which may be useful in tumors that have downregulated MHC in order to escape endogenous T-cell recognition.
Unlike T-cells, NK cells do not undergo clonal expansion. While this can effectively limit toxicity of the therapy, it also makes transduction and generation of suitable numbers for treatment a challenge. However, also unlike T-cells, there is an NK cell line that is effective in cell killing (NK-92), which can be transduced with a CAR as a clinical-grade product and used in patients. A clinical trial of anti-CD33 CAR NK-92 cells expressing the 4-1BB costimulation domain treated three AML patients, one of whom experienced MRD+ remission for a short period. Other sources for CAR NK cells that can be more readily expanded include cord blood, human embryonic stem cells, or induced pluripotent stem cells.
Additional strategies such as co-expressing IL-15 with the CAR have resulted in prolonged CAR NK survival. In 11 patients treated with IL-15 expressing CAR NK cells derived from HLA-mismatched cord blood, no GVHD or other major toxicity was observed, and 7 of the 11 patients had a CR (4 with lymphoma and 3 with CLL). Although NK cells generally survive an average of 2 weeks in humans, the cord blood-derived CAR NK cells utilized in this study expanded and persisted at low detectable levels for at least a year.
Like T-cells, NK cell activity can be modulated by a variety of cytokines. NK cells cultured with IL-12, IL-15, and IL-18 display enhanced effector function in response to further cytokines or tumor targets for weeks after the initial preactivation, developing a “memory-like” (ML) phenotype. Modifying these ML NK cells with a CAR endows them with the capacity to specifically recognize tumor in an antigen-dependent manner, and to control lymphoma burden in mice. Clinical trials using this strategy have yet to be performed.
Although many CAR NK studies utilize a traditional CAR design from a CAR T-cell, NK cells have different natural receptor requirements and the optimal costimulatory domain for NK cells may be different. Changing the costimulatory domain to the NK-cell-associated activating receptor 2B4, for example, results in increased engraftment and CAR NK persistence in mice. Additional NK costimulatory CAR domains are under investigation and further improved combinations are likely to continue to emerge.
