Single-cell analysis illuminates the gradients of immune cell functional states within human melanoma tumors, and facilitates characterization of tumor-reactive T-cells

Checkpoint blockade therapies that aim to reactivate antitumor immune responses have revolutionized cancer treatment, resulting in durable responses in a significant proportion of patients with advanced tumor progression. Nevertheless, many patients fail to reach long-term clinical benefit due to lack of response or acquired resistance. Inconsistency in therapy outcomes may be explained in part by recent findings suggesting that immune cell infiltrates in tumors are highly heterogeneous among patients. Therefore, comprehensive characterization of the diverse functional states exhibited by immune cell infiltrates is critical for the development of more effective immunotherapies. Here, we characterized immune cell infiltrates within tumors derived from 28 metastatic melanoma patients by single-cell RNA-seq of ~100,000 immune cells, and parallel T-cell receptor (TCR) sequencing, thereby generating an unbiased map of the expression signatures of immune cells, as well as clonality of T-cells within and between metastases. We identified within the tumor infiltrates naive, semi- and fully-activated effector, dysfunctional, and regulatory T-cells, as well as NK cells, and various myeloid subsets. While various immune cell types and cellular states are shared among patients, their frequency in each is highly heterogeneous even among similar tumor progression stages and treatment background. We noticed that clonally expanded T-cells predominantly adapted similar expression profiles. The frequencies of certain sub-populations were found to be correlated; notably, dysfunctional CD8 T-cell states were associated with prevalence of regulatory T-cells and follicular helper cells. The high-resolution map demonstrated gradient transitions between activation and dysfunctional states of CD8 T-cells, as well as variability within regulatory T-cell states. We used computational modeling to find the key transcription factors driving these gradients of expression states. Our analysis also puts forward novel candidate genes, which are highly correlate with known checkpoint targets within specific single-cell clusters, and may prove to be effective targets for checkpoint blockade. In order to validate our results, we performed in vitro reactivity assays (for available samples), which allowed us to determine autologous-tumor reactive and nonreactive TCRs, adding the reactive potential of T-cells as another layer of information to the single-cell expression signatures, facilitating inference of the functionality of clonally expanded T-cell populations in the tumor. Linking autologous tumor reactivity with single-cell expression profiles, we found that T-cells presenting reactive capability in vitro show a highly dysfunctional signature in the original tumor. The lack or presence of such reactive CD8 T-cells was also associated with patient response to immunotherapy (n=10). In conclusion, we present an atlas of immune cell infiltrates of human melanoma, revealing intra- and inter-patient heterogeneity of tumor immune infiltrates, highlighting regulatory circuits underlying different immune populations and their interactions. Our findings demonstrate how single-cell analysis may serve in the future as a tool for predicting response to therapy, provide rapid and effective tumor-immune state characterization, and ultimately lead to optimization of personalized immunotherapy. Citation Format: do Yofe, Hanjie Li, Anne van der Leun, Yaniv Lubling, Dikla Gelbard, Alexander C. J. van Akkooi, Amos Tanay, Ton N.M. Schumacher, Ido Amit. Single-cell analysis illuminates the gradients of immune cell functional states within human melanoma tumors, and facilitates characterization of tumor-reactive T-cells [abstract]. In: Proceedings of the Fourth CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; Sept 30-Oct 3, 2018; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2019;7(2 Suppl):Abstract nr B054.