pheno-seq – linking morphological features to gene expression in 3D cell culture systems

3D-culture systems have advanced cancer modeling by reflecting physiological characteristics of in-vivo tissues, but our understanding of functional intratumor heterogeneity including visual phenotypes and underlying gene expression is still limited. Single-cell RNA-sequencing is the method of choice to dissect transcriptional tumor cell heterogeneity in an unbiased way, but this approach is limited in correlating gene expression with contextual cellular phenotypes. To link morphological features and gene expression in 3D-culture systems, we present ‘pheno-seq’ for integrated high-throughput imaging and transcriptomic profiling of clonal tumor spheroids. Specifically, we identify characteristic EMT expression signatures that are associated with invasive growth behavior in a 3D breast cancer model. Additionally, pheno-seq determined transcriptional programs containing lineage-specific markers that can be linked to heterogeneous proliferative capacity in a patient-derived 3D model of colorectal cancer. Finally, we provide evidence that pheno-seq identifies morphology-specific genes that are missed by scRNA-seq and inferred single-cell regulatory states without acquiring additional single cell expression profiles. We anticipate that directly linking molecular features with patho-phenotypes of cancer cells will improve the understanding of intratumor heterogeneity and consequently be useful for translational research.