Heterotypic CD8 T cell clusters isolated from clinical samples are biologically distinct and enriched for antitumor activity

Emerging evidence suggests a correlation between CD8+ T cell-tumor cell proximity and immunotherapy response. However, it is unknown whether these cells can be captured as functional clusters from clinical samples. In defined human co-cultures, tumor antigen-recognizing T cells outcompeted unmatched T cells in forming clusters with tumor cells, prompting us to investigate whether this feature could be used to isolate tumor-reactive T cells directly from cancer specimens. By conventional and imaging flow cytometry, we show here that from 21/21 human melanoma metastases, we were able to isolate heterotypic clusters, comprising CD8+ T cells interacting with one or more tumor cells and/or antigen-presenting cells (APCs). Single cell RNA-sequencing revealed that CD8+ T cells from clusters were enriched for tumor-reactive and exhausted gene signatures. Integration with T cell receptor (TCR)-sequencing showed increased clonality of clustered T cells, indicative of expansion. In-depth analyses revealed that these T cells had conjugated with tumor cells and various APCs, each exhibiting specific enriched cell states, which were linked to distinct patterns of cell-cell communication. CD8+ T cells expanded from clusters ex vivo exerted on average 9-fold increased killing activity towards autologous melanomas, accompanied by enhanced cytokine production. Also upon adoptive cell transfer (ACT) into mice, T cells from clusters showed superior patient-derived xenograft (PDX) killing associated with more T cell infiltration and activation. Together, these results demonstrate that tumor-reactive CD8+ T cells are enriched in functional clusters with tumor cells and/or APCs, and that they can be isolated and expanded from clinical samples. Typically excluded during single-cell sorting by flow cytometry, these distinct heterotypic CD8+ T cell clusters serve as a valuable source amenable to deciphering functional tumor-immune cell interactions, while they may also be therapeutically explored.

• Type: Transcriptome Sequencing
• Archiver: European Genome-Phenome Archive (EGA)