Defining the functional and spatial properties of tumor-specific tumor-infiltrating CD8 T cells

Successful anti-tumor T cell responses rely on the interplay between T cell receptor (TCR) specificity, T cell phenotype, and cellular interactions within the tumor microenvironment. Despite recent advances targeting each of these axes, many tumors remain refractory to therapy, underscoring the need to understand how distinct solid tumors sculpt microenvironments that present unique challenges to T cells. Here, we describe comprehensive analyses into two solid tumor contexts, head and neck squamous cell carcinoma (HNSCC) and renal cell carcinoma (RCC), leveraging a high-throughput methodology to uncover not only the phenotypes of T cell subpopulations driving response to immunotherapy, but also their TCR specificities and cellular interactions underlying the native anti-tumor T cell response. To define how T cell phenotypes relate to therapeutic modulation of the native T cell response, we analyzed pre- and post-therapy biopsies from responders and non-responders to neoadjuvant immune checkpoint blockade in HNSCC. This enabled us to identify a population of pre-existing exhausted T cells, with a tissue-resident phenotype and expressing cytotoxic transcripts, which was enriched in responders across multiple cohorts, suggesting a conserved role in therapeutic response. While T cell phenotype provides important insight into therapeutic response, it does not reveal the antigenic drivers of tumor recognition. To address this, we developed a high-throughput platform to assess hundreds of TCRs against expansive libraries of immunogenic targets, including patient-derived tissues and antigens. This platform enabled us to investigate T cell antigen specificity in RCC, a tumor with few known T cell antigens and a weak correlation between T cell infiltration and immunotherapy response. Across six patients, we identified 72 tumor-specific TCRs recognizing multiple antigens, including a highly immunogenic neoantigen and a novel tumor-associated antigen shared by five patients. Spatial mapping revealed two opposing immune processes: tertiary lymphoid structures (TLS), aggregates of B and T cells within tumors associated with successful immunotherapy response, harbored tumor-specific and progenitor exhausted T cells, whereas macrophage-rich boundary regions sequestered terminally exhausted T cells. These findings support a model in which progenitor exhausted, anti-tumor T cells are maintained within TLS in RCC, but are prevented from controlling tumors by macrophage-rich immunosuppressive boundaries. Together, these studies define the cellular phenotypes underlying T cell responses across diverse solid tumors and establish a high-throughput approach for identifying tumor-specific TCRs. We identify exhausted T cells as the dominant tumor-specific population, with phenotypic heterogeneity in the form of cytotoxicity or progenitor exhaustion programs that likely contribute to variable patient responses. These findings could enable the rapid identification of tumor-specific TCRs for cellular therapies, and improve prediction of therapeutic efficacy based on pre-existing T cell populations within patient tumors.