Title: Reverse Sensitivity Analysis for Identifying Proteomics Signatures of Cancer Abstract Cancer is a complex disease in which genetic disruptions in cell signaling networks are known to play a significant role. A major aim of cancer systems biology is to build models that can predict the impact of these genetic disruptions to guide therapeutic interventions (i.e. personalized medicine). A prominent driver of cancer cell growth is signaling pathway deregulation from mutations in key regulatory nodes and loss/gain in gene copy number (CNV). However, current mathematical modeling approaches do not adequately capture the impact of these genetic changes. Reasons for this include the poorly understood layers of regulation between gene expression and protein activity, and limitations in most modeling and protein measurement technologies. In addition, there is a paucity of overarching hypotheses that can link specific gene expression or mutation patterns to the cancer phenotype. Recent work by our group has resolved some of the technical challenges that have hindered the application of proteomics technologies to cancer systems biology research. It has also suggested a new approach for using quantitative proteomics data to understand mechanisms driving cancer cell behavior. Using an ultrasensitive, targeted proteomics platform that can measure both abundance and phosphorylation of proteins present at only hundreds of copies per cell, we found that signaling pathways appeared to be controlled by only a limited number of key nodes whose activity is tightly regulated through low abundance and feedback phosphorylation. We propose to build on these findings by critically testing the hypothesis that CNV and genetic mutations dysregulate signaling pathways in cancer by shifting control from tightly regulated nodes to poorly regulated ones. This will be done by systematically identifying key regulatory nodes of normal and cancer cells using CRISPRa/i screens, determine the relationship between protein abundance and signaling pathway activities using ultrasensitive targeted proteomics and phosphoproteomics and then use these data to semi-automatically generate mathematical models of the functional topology of the signaling pathways. Specifically, we propose to: 1) Use targeted CRISPR gene perturbation libraries to identify the regulatory topologies of signaling pathways important in cancer and how they are disrupted by common cancer mutations, 2) Use the CRISPR perturbation and proteomics data to semi-automatically build predictive models of cancer cell signaling pathways, and 3) Combine modeling and perturbation screens to understand how feedback regulation in cancer contributes to drug resistance. This work will result in simplified, computationally tractable yet mechanistic models of signaling pathways and provide network maps of feedback and crosstalk circuits that can be used to rapidly map the regulatory state of cells. Most important, it will provide a generic platform for translating protein abundance and phosphorylation patterns into a “state” snapshot of cancers that can lead to predicting their response to specific drugs.

Narrative Cancer is an extremely complex disease which is frequently caused by signaling pathway deregulation from genetic mutation or loss/gain in gene copy number. Modern analytical techniques have provided a wealth of data on the molecular changes associated with cancer mutations, but it has been extremely difficult to use this information to design targeted therapies. We propose a new methodology that combines CRISPR-based targeted screens, advanced proteomics technologies, and a new mathematical modeling approach for identifying proteomics signatures of altered signaling pathways in cancer that can be used to build predictive models and explore mechanisms of drug resistance.