Single-cell analysis of tumor progression reveals the function, structure, and evolution of cancer archetypes

The classic cancer evolution model posits that driver mutations sweep the population sequentially as the complete set of hallmarks are assembled by the neoplastic clone. However, recent work has challenged this model revealing that most tumors contain highly complex dynamics with genetic diversity reflecting distinct clonal architectures. The functional and phenotypic heterogeneity also has been shown to have a crucial influence on the fate of the tumor5–10. However, it is not well understood how distinct tumor clonal populations coexist and function. Here, we study tumor architecture at the level of individual cells by sampling a zebrafish melanoma tumor over time and space. We found that cancer transcriptional programs can be classified to three archetypes, each exploiting the existing neural crest, mature melanocytes, and stress modules, and distinct intra-tumor locations. Strikingly, these archetypes are conserved in human melanoma. Further, we found that the cancer cells are comprised of two distinct clones, where one expresses a unique archetype. Over time, we found that the cells of this clone adapt by exhibiting a more similar profile to the corresponding archetype. Overall design: Single-cell RNA sequencing of zebrafish tumor cells from 2 zebrafish at multiple time points.