Engineering and preclinical evaluation of CXCR3-expressing, EGFR-targeted CAR-NK cells for tumor homing in non–small cell lung cancer and esophageal squamous cell carcinoma models.

Abstract

e14515 Background: CAR-NK cell therapy for solid tumors is hindered by insufficient tumor infiltration. To address this limitation, we engineered anti-EGFR CAR-NK cells, which is overexpressed in non-small cell lung cancer and esophageal squamous cell carcinoma (ESCC)—to co-express the chemokine receptor CXCR3. This strategy aims to enhance homing to tumor microenvironments enriched with CXCR3 ligands CXCL9 and CXCL10. Methods: Third-generation anti-EGFR CAR constructs—either fused with CXCR3 via a P2A self-cleaving peptide (CAR-CXCR3) or co-expressed with IL-15 as a control (CAR-IL-15)—were generated. Primary human NK cells were lentivirally transduced and comprehensively characterized using: flow cytometry to assess CAR and CXCR3 expression and cellular purity; Annexin V/propidium iodide (PI)–based cytotoxicity assays against EGFR-positive A549 (NSCLC) and KYSE150 (ESCC) target cells; Transwell migration assays toward CXCL9 and CXCL10, followed by functional assessment of post-migration cytotoxicity; quantification of cytokine secretion (IFN-γ, granzyme B) and CD107a surface exposure (a marker of degranulation) upon co-culture with tumor cells; and in vivo evaluation of antitumor efficacy in NSG mice bearing subcutaneous KYSE150 tumors engineered to overexpress CXCL9/10. Mice received three intravenous doses (1 × 10⁷ cells per dose) of CAR-CXCR3-NK cells or phosphate-buffered saline (PBS) as control. Results: CAR-NK cells achieved ~50% transduction efficiency and > 93% purity. CXCR3 expression was robustly and selectively upregulated only in CAR-CXCR3-NK cells. Both CAR-NK variants exhibited potent, EGFR-dependent cytotoxicity significantly exceeding that of unmodified NK cells. CAR-CXCR3-NK cells demonstrated markedly enhanced migration toward CXCL9 and CXCL10 (68.2% vs. 32.5% migrated cells; p < 0.01), an effect fully abrogated by CXCR3 blockade, and retained strong cytolytic activity following migration. Upon engagement with tumor cells, CAR-CXCR3-NK cells secreted significantly higher levels of IFN-γ and granzyme B (p < 0.05) and displayed increased CD107a degranulation (p < 0.01). In vivo, CAR-CXCR3-NK treatment led to significant suppression of tumor growth (p < 0.001); CAR-NK cells were readily detectable in peripheral blood, and a trend toward prolonged survival was observed (median survival: 18.2 vs. 9.5 weeks; p = 0.06). No unexpected or off-target toxicities were noted. Conclusions: EGFR-directed CAR-NK cells co-expressing CXCR3 demonstrate enhanced chemotaxis toward CXCL9/10, superior in vitro cytotoxic function, and robust in vivo antitumor efficacy. Incorporating CXCR3 represents a promising and rationally designed strategy to overcome the critical barrier of poor tumor infiltration in CAR-NK therapy for EGFR-overexpressing solid malignancies.