Metabolic landscape of the tumor microenvironment

The tumor milieu consists of numerous cell types with each cell existing in a different nutritional environment. However, a characterization of intratumoral metabolic heterogeneity at the single-cell level in human cancer is not established. Here, we develop a computational pipeline to analyze metabolic gene expression programs of single cells within human tumors. In two representative cancer types, melanoma and head and neck, we apply this algorithm to define the single-cell metabolic landscape of human tumors. We find that malignant cells in general have higher metabolic activity and higher metabolic variation than previously observed from studies of bulk tumor comparisons. Indeed, most of the observed metabolic variation of single tumor and normal cells were found to be inconsistent with comparisons with bulk tumor samples. Variation in the expression of mitochondrial programs is the major contributor to intratumoral metabolic heterogeneity. Surprisingly, the expression of both glycolytic and mitochondrial programs strongly correlates with hypoxia in almost all cell types. Immune and stromal cells could also be distinguished by their metabolic features. Taken together this analysis establishes a computational framework for characterizing metabolism using single cell expression data and defines principles of intratumoral metabolic heterogeneity.