Introduction

Chimeric antigen receptor (CAR) expressing natural killer (NK) cell-based immunotherapy is offering a paradigm shift in personalized cancer medicine as they can be utilized as universal cell‐based therapy without requiring human leukocyte antigen (HLA) matching like T cells. Though NK cell based clinical trials are ongoing, identification of novel targets is necessary to achieve better efficacy. Human leukocyte antigen G (HLA-G) is a central protein that plays a critical role in providing immune tolerance to the fetus in pregnant women. As a class I major histocompatibility complex (MHC) protein, HLA-G suppresses the function of NK cells and a subset of T cells via engagement with inhibitory receptors such as leukocyte Ig-like receptor subfamily B member 1 (LILRB1) and member 2 (LILRB2), and the killer immunoglobulin-like receptor 2DL4 (KIR2DL4). Cancer cells aberrantly upregulate HLA-G on their cell surface, thus, inducing immunosuppression within the tumor-microenvironment. This makes HLA-G an attractive cell therapy target. Here, we propose to develop CAR-NK cells targeting HLA-G protein as a novel strategy for cancer immunotherapy, using NK cells from healthy human donor cord blood.

Methods

Flow cytometry-based cell surface staining of HLA-G was done on different cancer cell lines of acute myeloid leukemia (AML), glioblastoma and pancreatic cancer. The mRNA expression analysis of HLA-G isoforms was done on the above-mentioned cancer types using TCGA data set. NK cells were isolated from healthy human donor cord blood via a negative selection kit. After days of expansion with specific feeder layer of cells, transduction was performed with a retroviral construct containing the ScFv sequence derived from a specific human anti-HLA-G monoclonal antibody. In addition, our construct incorporated the gene for interleukin 15 (IL-15), an important cytokine for NK proliferation and persistence in vivo. To determine the role of different costimulatory molecules on HLA-G CAR NK potency, we compared two different signaling domains coming from CD28 and DAP10. CAR-NK cells in vitro functionality was tested by co-culturing with tumor cell lines and further assessing cell death via incucyte and xcelligence instrument. Flow cytometry was performed to measure NK cytotoxic markers e.g., CD107a, TNF-a, IFN-g.

Results

We observed elevated HLA-G surface expression on cell lines from AML, glioblastoma, pancreatic ductal adenocarcinoma, breast cancer and renal cancer. HLA-G isoform mRNA expression analysis from TCGA data also showed differential elevated mRNA expression of HLA-G isoforms on those cancer types.

We obtained an average of 50-70% transduction efficiency and HLA-G CAR expression on cord blood derived NK cells. In short time co-culture experiments, HLA-G CAR NK cells killed OCI-AML3 cells even in the absence of IL-15. HLA-G CAR NK cells also showed cytotoxicity towards GSC20 spheroids and PANC1 cells. When HLA-G CAR NK cells were incubated with plate bound HLA-G protein, they showed strong activation via upregulation of CD107a, TNF-a, IFN-g compared to the non-specific protein CD5. To understand the functionality better, we generated CAR construct with CD28 and DAP10 costimulatory domain. Both costimulatory domains containing constructs also showed robust cytotoxicity against OCI-AML3 cells at short-term as well as long-term tumor rechallenge killing assay.

Conclusion

Our finding suggests that HLA-G is upregulated at protein level in different hematological and solid cancer models and that HLA-G targeting CAR NK cells, generated from human donor cord blood, could be a novel off-the-shelf cellular therapy approach for different malignancies.

Tumor xenograft experiments to study the in vivo safety and efficacy of HLA-G targeting CAR NK cells are under way.