Chimeric antigen receptor (CAR)-modified natural killer (NK) cells represent a promising immunotherapeutic modality for cancer treatment. However, their potential utilities have not been explored in hepatocellular carcinoma (HCC). Glypian-3 (GPC3) is a rational immunotherapeutic target for HCC. In this study, we developed GPC3-specific NK cells and explored their potential in the treatment of HCC. The NK-92/9.28.z cell line was established by engineering NK-92, a highly cytotoxic NK cell line with second-generation GPC3-specific CAR. Exposure of GPC3 HCC cells to this engineered cell line resulted in significant in vitro cytotoxicity and cytokine production. In addition, soluble GPC3 and TGF-β did not significantly inhibit the cytotoxicity of NK-92/9.28.z cells in vitro, and no significant difference in anti-tumor activities was observed in hypoxic (1%) conditions. Potent anti-tumor activities of NK-92/9.28.z cells were observed in multiple HCC xenografts with both high and low GPC3 expression, but not in those without GPC3 expression. Obvious infiltration of NK-92/9.28.z cells, decreased tumor proliferation, and increased tumor apoptosis were observed in the GPC3 HCC xenografts. Similarly, efficient retargeting on primary NK cells was achieved. These results justified clinical translation of this GPC3-specific, NK cell-based therapeutic as a novel treatment option for patients with GPC3 HCC.

Our previous study indicated that GPC3-targeted chimeric antigen receptor (CAR) T cell therapy has a high safety profile in patients with hepatocellular carcinoma (HCC). However, the response rate requires further improvement. Here, we analyzed the combined effect of GPC3-CAR T cells and sorafenib in both immunocompetent and immunodeficient mouse models of hepatocellular carcinoma. In immunocompetent mouse model, mouse CAR (mCAR) T cells induced regression of small tumors (approximately 130 mm 3 tumor volume) but had no effect on large, established tumors (approximately 400 mm 3 tumor volume). Sorafenib, at a subpharmacologic but not a pharmacologic dose, augmented the antitumor effects of mCAR T cells, in part by promoting IL12 secretion in tumorassociated macrophages (TAMs) and cancer cell apoptosis. In an immunodeficient mouse model, both subpharmacologic and pharmacologic doses of sorafenib had limited impacts on the function of human CAR (huCAR) T cells in vitro and showed synergistic effects with huCAR T cells in vivo, which can at least partially be ascribed to the upregulated tumor cell apoptosis induced by the combined treatment. Thus, this study applied two of the most commonly used mouse models for CAR T cell research and demonstrated the clinical potential of combining sorafenib with GPC3-targeted CAR T cells against HCC.