Integrated combinatorial functional genomics and spatial transcriptomics of tumors decodes genotype to phenotype relationships

Linking the complex genetic changes underlying cancer to relevant disease-phenotypes poses a challenge. Therefore, we present CHOCOLAT-G2P, a scalable approach that integrates multiplex in vivo functional genomics with spatial transcriptomics. By redeploying RNA-templated ligation probes of commercial spatial transcriptomics technology, we streamline mapping composite genetic alterations and transcriptome-wide phenotyping on the same tissue section on a single readout platform. Using this framework, we studied combinatorial effects of 8 perturbations that induce autochthonous mosaic liver tumors sampled from 256 genotypes. Interrogating 324 tumors across six [~]6x6 mm2 sections, we charted phenotypic landscapes of genotypically-defined tumor ecosystems, revealing zonation-associated hepatocellular carcinoma subclasses and associations between tumor subtypes and stromal-as well as immune-cell signatures. Further, we decoded epistasis within compound genotypes uncovering opposing roles of Vegfa and mutant Ctnnb1 to cholangiocarcinoma development. Thus, CHOCOLAT-G2P lays a foundation to decipher how combinations of alterations interact to reprogram tumor cells and their microenvironment within the holistic context of tissue and whole organisms. (https://chocolat-g2p.dkfz.de/).