FAP-Targeted LTβR Agonist Drives HEV Differentiation and Immune Niche Formation for Improved Immunotherapy Response in Solid Tumors

PURPOSE: Immune checkpoint inhibitors (CPI) have revolutionized cancer therapy, yet many patients derive limited benefit because of poor immune infiltration within the tumor microenvironment (TME). The presence of tertiary lymphoid structures (TLS) and high endothelial venules (HEV) correlates with improved immunotherapy responses. This study evaluated whether selective activation of lymphotoxin β receptor (LTBR) signaling, targeted to fibroblast activation protein (FAP)-expressing tumor stroma, could remodel the TME to enhance immune cell infiltration and potentiate immunotherapy. EXPERIMENTAL DESIGN: FAP-LTBR was engineered as a novel tumor-targeted LTBR agonist and was characterized for its binding, activation, and immunomodulatory properties in vitro and in vivo. Functional effects were assessed using primary human endothelial cells, three-dimensional (3D) microfluidic vascular models, and multiple murine tumor models. Spatial transcriptomics and 3D immunophenotyping elucidated TME remodeling. Therapeutic efficacy was tested as monotherapy and in combination with CPIs or T-cell engagers. RESULTS: FAP-LTBR selectively activated endothelial cells and induced chemokine secretion in a FAP-dependent manner, enhancing T-cell adhesion and extravasation in vitro. In murine models, FAP-LTBR promoted HEV differentiation, TLS-like immune aggregates, and broad tumor inflammation marked by increased B- and T-cell infiltration, including stem-like TCF1+ CD8+ T cells. FAP-LTBR synergized with CPIs and T-cell engagers to induce durable tumor regression, with superior CD8+ T-cell recruitment and redistribution into tumor cores. CONCLUSIONS: FAP-LTBR represents a first-in-class tumor-targeted LTBR agonist that remodels the TME, promoting HEV differentiation, immune cell infiltration, and the formation of organized lymphoid aggregates, which collectively enhance immunotherapy efficacy in preclinical models.