Towards a complete characterization of the metastasis founder clones in colorectal cancer - The mechanistic basis of metastasis formation in humans remains elusive. We do not know whether metastases are formed by random disseminated cells, or whether molecularly defined metastatic clones with superior ability to colonize specific organs exist. In metastasis experiments in mice, (rare) subpopulations of cells with increased metastatic potential have been shown to exist. This potential is mitotically heritable and associated with specific gene expression programs. In humans, similarly compelling data does not exist, and a unified theory of metastasis that harmonizes observations in animals and humans and makes testable predictions with clinical relevance is still missing. Using phylogenetic analysis of hundreds of human colorectal cancer samples, we have recently shown that anatomically distinct distant metastases are typically formed by only one primary tumor subpopulation (“metastasis founder clone”). Here, we propose a novel experimental design that aims to identify the molecular traits of metastasis founder clones in humans. In specific aim 1, we will conduct a rigorous search for recurrent gene expression patterns in metastasis founder clones. We will collect grid biopsies from newly resected primary colorectal cancers and matched synchronous liver metastases, reconstruct evolutionary trees from microsatellite mutation data and precisely annotate metastasis founder areas vis-à-vis the remainder of the primary tumor (“bystander areas”). Then, we will perform RNA sequencing to identify founder-specific gene expression patterns across 70 patients, with the goal of defining a universal “founder signature” that identifies cell populations with superior metastatic ability within a primary tumor. In specific aim 2, we will investigate mitotically heritable molecular alterations that could potentially give rise to a metastasis-enabling gene expression signature. We prioritize somatic copy number alterations (SCNAs) and DNA methylation patterns as candidates for such alterations, but to be comprehensive, we will also evaluate driver mutations. We will perform shallow whole genome, exome and reduced representation bisulfite sequencing in founders, selected bystanders, and matched metastases to determine whether founders are enriched for a specific subset of SCNAs, driver mutations and/or methylation changes. In specific aim 3, we will validate and mechanistically explore the biological properties of metastasis founder clones with xenotransplanation assays of patient-derived organoids (PDOs). In Aim 3A, we will deploy PDO technology to test the hypothesis that PDOs from metastasis founder areas are more likely to metastasize to the mouse liver than PDOs derived from bystander areas, providing an independent method for corroborating our human results. In Aim 3B, we will use the PDO xenotransplantation system to explore the biological function of genes highlighted as metastasis-relevant in aims 1 and 2 via loss- and gain-of-function genetic screens. This in vivo experimental framework will enable in depth mechanistic follow-up on leads gained in Aims 1 and 2. In total, this proposal outlines the first steps toward the urgent clinical imperative of identifying the molecular features of metastasis founder clones in colorectal cancer.
