<p>Immune checkpoint blockade (ICB) has shown limited activity in HER2⁺ breast cancer, yet the mechanisms underlying this refractoriness remain unclear. Integrating single-cell transcriptomics from untreated human and murine HER2⁺ tumors with an anti-HER2 non-sensitive mouse model, a neoadjuvant non-pCR patient cohort, functional co-culture assays, and in vivo perturbations, we identify TIGIT signaling from malignant cells (via CD112) to CD8⁺T cells as a dominant immunosuppressive axis. High TIGIT and CD112 expression correlate with poor clinical outcomes and with the enrichment of TIGIT⁺CD8⁺T cells after anti-HER2 therapy. Therapeutically, combining anti-TIGIT with anti-HER2 reprograms the tumor microenvironment, expanding activated CD8⁺ T cells with enhanced effector function, increasing IFN-γ production, and restoring MHC-I on tumor cells. A central mechanistic node is the reinstatement of the costimulatory receptor CD226 on CD8⁺T cells. TIGIT blockade induces CD226, and CD8 dependence is required for efficacy. Neutralizing CD226 abrogates cytotoxicity and IFN-γ secretion. Multiplex tissue analyses further show that intratumoral CD8⁺CD226⁺T-cell density predicts improved overall and disease-free survival in HER2⁺ disease. Collectively, these data reveal that restoring CD226-mediated co-stimulation overcomes TIGIT-refractory immunity and sensitizes HER2⁺ tumors to anti-HER2 therapy, positioning CD226 as both a pharmacodynamic driver and a clinically actionable biomarker for patient selection and response monitoring.</p><p></p>