Predicting Cancer Cell Response to Endogenous and Exogenous Perturbations at the Single Cell Level

点击翻译按钮获取中文摘要

The study of cancer has been predicated on the discovery of individual events, mechanisms, and processes that are implicated in the malignant transformation of normal cells. The expectation has been that learning how tumors arise would teach us how to defeat them, by providing pharmacologically actionable targets. Unfortunately, while the etiology of many cancers has been elucidated in painstaking detail, curative therapies for the more aggressive subtypes remain elusive. Indeed, once their regulatory and signaling logic is dysregulated by events leading to transformation, cells are no longer constrained to well-defined physiologic states but rather transition to a novel dysregulated and highly plastic landscape, where de-differentiation and trans-differentiation events become common and novel synergies with other non-transformed cell populations are forged, leading to the emergence of bona fide neomorphic organs. Thus, to study cancer, we must generate cellular network models that are fully capable of recapitulating dynamic cell behavior and response to drug perturbations, mutations, and interactions with cells in the tumor microenvironment (TME). Our prior research suggests that identifying vulnerabilities that are more universal and less likely to be defeated by the cancer cell’s remarkable adaptive nature will require a quantum leap in our ability to dissect and interrogate intra- and inter- cellular network models that are predictive of the dynamic behavior of cancer as a bona fide neomorphic organ. To address this challenge, we will create the first generation of genome- and proteome-wide network models that can effectively predict the probabilistic, time-dependent response of mammalian cells to small molecule and genetic perturbations, as well as their ability to plastically reprogram across the relatively small number of molecularly distinct states detected in virtually all human malignancies. To achieve these goals, we will leverage a repertoire of state-of-the-art experimental and computational advances, developed over the last six years with R35 funding, for the creation of systematic, large-scale Transcriptional Regulator Knock-down (TREK) single cell profiles at multiple time points following CRISPRi-mediated silencing of regulatory proteins. Owing to their novelty and value, these reagents have already been distributed by Addgene to >270 labs. TREK data, comprising hundreds of thousands to millions of individual molecular profiles and billions of individual molecular readouts, will be used for causal network learning with intervention, allowing the assembly of probabilistic models of cell regulation that effectively recapitulate the dynamic behavior of both normal and cancer-related cells, as well as their interactions within the TME. By the end of the R35 funding cycle, we expect to be able to infer the time-dependent activity of most regulatory and signaling proteins following arbitrary genetic or pharmacologic perturbations, and to infer which proteins were affected by a perturbation based on generation of a small (n ≥ 6) longitudinal series of scRNA-seq profiles. These new cellular network models will also allow rapid, systematic elucidation of mechanisms that have been traditionally studied via low-throughput, hypothesis-driven assays.