In situ tumor arrays reveal early environmental control of cancer immunity [NGS2751]

The immune phenotype of a tumor is a key predictor of its response to immunotherapy. Patients who respond to immune checkpoint blockade generally present with tumors infiltrated by T cells, a phenotype referred to as ‘inflamed’. However, not all inflamed tumors respond to therapy, and even lower response rates occur among patients with tumors that lack T cells (‘desert’) or that spatially exclude T cells to the periphery of the tumor lesion (‘excluded’). Despite the importance of these tumor immune phenotypes in patients, little is known about their development, heterogeneity or dynamics due to the technical difficulty of modeling and tracking these features in situ. Here, we introduce STAMP (skin tumor array by microporation), a preclinical approach that combines in vivo high-throughput time-lapse imaging with next generation sequencing of tumor arrays.  Using this approach, we follow the early formation of thousands of tumors to show that development of a given immune phenotype is not under strict control of tumor genetics or transcriptional state, but is also influenced by local features of the tumor microenvironment. Only the spatial organization of T cells, specifically early infiltration of T cells recruited by fibroblasts and monocytes into the core of a tumor, was predictive of functional T cell attack and tumor rejection. Evaluating the dynamic immune-history of tumors revealed that early conversion to the immune inflamed phenotype was predictive of therapy-induced or spontaneous tumor regression. Thus, STAMP captures the dynamic and complex relationships of spatial, cellular and molecular components of tumor progression and has the potential to translate therapeutic concepts into successful clinical strategies. KPP-GFP (C57Bl/6) cells were implanted into the ears of B6.129S6-Rag2tm1Fwa N12 mice (C57Bl/6NTac background) that were adoptively transferred with 4x10^6 naive CD4tdTomato-positive Tcells (B6 N/J background) the same day. The STAMP microtumors were imaged daily starting at day 3 after tumor cell implantation. On day 8 after tumor cell implantation tumors were classified as either (1) Inflamed, (2) Excluded, (3) Ignored/Desert or (4) Resolved. Thereafter a pool of 3-4 tumors (ranging from from a size of 0.7 - 1.2 um diameter) of each phenotype (including small amounts of surrounding healthy stroma) were biopsied from the two ears of one single mouse and enzymatically digested (in the presence of Actinomycin D) to generate single cell suspensions. The cells were then incubated in a staining solution containing Calcein Blue and 7AAD and enriched for live and non-dead cells by flow cytometry. The viability of single cells suspensions as determined by Vicell were (1) 59.3%, (2) 71.4%, (3) 90.6% and (4) 80%. The cell number ranged from 5-6x10^4 for (3) and (4) to about 10-11x10^4 for (1) and (2) as determined by flow cytometry and Vicell. Comparing 5' single cell RNAseq data between the four STAMP microtumor immune phenotypes should provide insight into differences of the cellular compositions of their tumor-immune and -stromal microenviornments, reveal differences of the associated TCR repertoire, help in the generation of transcriptome-based classifiers of the different immune phenotypes and lead to the discovery of important and or novel pathways associated with the different immune phenotypes.