(2aw) Developing Computational Tools for in silico elucidation of Cancer Mechanisms, Microenvironment, and Drug Repositioning Candidates

The overarching goal of my proposed research program is the development and application of various novel optimization-based tools for model-based analysis and phenotype prediction for cancer biology and treatment applications with three research foci: i) novel tool and model-type development, particularly incorporating signal pathways and epigenetics; ii) quantitative study of cancer biology focusing on biomarker mechanisms, the tumor microenvironment, and their interdependence; and iii) identification of drug repositioning candidates. This will be accomplished through three inter-related projects: Project 1: novel in silico analysis tools and novel model-type development to include often overlook network components such as epigenetics, cell logic, and signaling pathways; Project 2: identification and analysis of underlying mechanisms of cancer biomarkers, modeling of the tumor microenvironment, and their interdependence; and Project 3: the incorporation of drug-interaction networks into cancer models for identifying drug repositioning candidates. These projects will be done in collaboration with other laboratories for the refinement of models by the in vivo and in vitro testing of model prediction. Figure 1 shows an overview of proposed research projects. Funding for these projects will come largely through the National Institutes of Health (NIH) as this program will align with cancer biology and treatment research areas of the National Cancer Institute (NCI). Project 1 may additionally be funded by the Systems and Synthetic Biology (SSB) cluster under the Division of Molecular and Cellular Biosciences Core Programs (MCB).

Project 1: Novel tool and model-type development incorporating signal pathways and epigenetics

Motivation: Advances in the ability to explore biological systems including fast and cheap genome sequencing, the ‘omics’ revolution, single-cell ‘omics’, and personalized medicine (particularly cancer treatment) based on individual biomarkers has and will continue to produce a wealth of in vivo biological data. This data can inform metabolic and network models which offer quicker and cheaper alternatives to traditional in vivo and in vitro analysis techniques to elucidate underlying mechanisms or emergent network properties arising from such data. At present, these computational analysis tools are limited, often applying data from only one or two omics levels; ignoring or strongly simplifying regulation; or ignoring omics data which do not directly act upon metabolic enzymes. This can be problematic for studying cancer, as signal pathways are key to cancer biology, strongly affect cell metabolism, and have been identified as promising therapeutic targets. Further, epigenetics are often dysregulated in cancer and key to initiation, progression and metastasis. As such new modeling tools and types will be created to integrate these additional cell network layers. By creating a suite of tools and a new signal- and epigenetics-incorporating metabolic model type, the accuracy and predictiveness can be greatly enhanced, and could lead to patient-specific omics models of disease for personalized therapy.

Objective: The proposed project aims to develop new in silico analysis tools and models applicable to cell-and community-level flow of genetic information, regulation, and signaling. Through this project, my group will i) produce a suite of more sophisticated systems biology tools which can use the current wealth of multi-omics data, particularly with respect to cancer biomarkers; and ii) create a new metabolic model type or types which incorporate the effects of signaling pathways and genes which do not directly produce or affect a metabolic gene.

Project 2: Quantitative study of cancer biology focusing on biomarker mechanisms, the tumor microenvironment, and their interdependence

Motivation: Cancer refers to a set of more than 100 related genetic disease which arise due to genetic mutations which cause cells to divide uncontrollably, damaging the tissue in which they arose and/or damaging other tissues throughout the body and is the leading cause of death worldwide. As cancer arises from genetic mutations, different cancers have different biomarkers, biological signatures of disease type, prognosis, and mechanism, which are often used to inform patient treatment. While some biomarkers are well-studied, such as breast cancer markers Estrogen Receptors (ER), HER2neu (HER2), and Breast Cancer Gene 1 (BRCA1), the omics revolution has resulted in an explosion of new biomarkers or patterns of biomarkers including pipelines for biomarker discovery, collectively outpacing in vivo analysis of mechanisms of biomarkers. Many biomarker resources exist, and a generally very recent, examples include The Cancer Genome Atlas (TCGA, 2006), CBD (2018), dBMHCC (2020), OncoMX (2020), the signal web tool (2020), and MarkerDB (2021). Therefore, there is a wealth of clinically relevant biomarkers for many of which we lack and understanding of the underlying biological mechanism due to their only recent collation or discovery. Further, biomarkers alone do not explain cancer biology and behavior, and integration of the cellular microenvironment though considering interactions between different cell types are crucial to understanding the ecology and evolution of cancer cells within a host. Mathematical and computational tools are crucial as a relatively low-cost method of exploring the role of environment upon cancer cells. Recent advances in both human cellular modeling (with the Virtual Metabolic Human in 2019) and whole-body human modeling (in 2020, modeling more than two dozen interconnected tissue types in personalized sex-specific models) have resulted in highly detailed genome-scale metabolic models. Further sophisticating these models into new modeling types with new tools proposed by Project 1 can render them more fit for the study of cancer biology.

Objective: This project will apply the computational analysis tools developed in Project 1 to i) reconstruct more sophisticated models of health cell metabolism, signaling, and epigenetics; ii) construct cancer cell model from the healthy model using biomarker databases, hypothesize marker effects on the cancer cell through modeling iii) quantify and study the interactions of cancerous cells with its microenvironment, both with healthy host cells, and other cancerous cells; and iv) identify interdependence between biomarkers and the cancer microenvironment, e.g. biomarkers that strongly affect microenvironment interactions or conditions that make biomarkers more viable.

Project 3: Identification of drug repositioning candidates

Motivation: For at least two decades, metabolic modeling have been considered potentially effective in drug discovery and repositioning as it can span multiple scales and leverage cell complexity to drive leading to hypothesis-driven discovery, though genome-scale models have not been used in repurposing drugs for cancer treatment until recently, with the field still being considered new. Recent studies have screened and proposed drug repositioning candidates for prostate, breast, and colorectal cancers (all in 2019) through gene and metabolite essentiality analyses using genome-scale models. These however are limited to drugs acting directly on the metabolic network and metabolic enzymes. To predict other on- and off-target effects of drugs on a genome-scale network, machine-learning-based predictive tools for Drug-Target Interactions (DTIs), developed to aid in drug discovery and repositioning, can be used. Combining more sophisticated human metabolic network models and tools (as proposed in Project 2) with predictive tools for DTIs allows for in silico prediction of drug candidate beyond direct metabolism effectors. This project proposes to use the detailed human stoichiometric model with additional signaling and epigenetic modelling layers to perform network level drug repositioning analyses to determine drug candidate efficacy.

Objective: This project will i) Develop Drug-Target Interaction (DTI) networks for cancer treatment drug repositioning candidates; ii) create a framework for the integration of DTI networks into metabolic models; and iii) determine the suitability and efficacity of repositioning candidates based on model-predicted metabolic shifts due to drug introduction and the drugs’ differential effects on cancerous and healthy cells.

Teaching Interests

My current views of teaching and learning have been influenced by my own experiences as a student (what has worked for me) and teacher (what has worked for others) and my knowledge of pedagogy. Generally, my views and practices align with Social Cognitive Theory and Constructivist Theory. Social Cognitive Theory states that learning occurs through observing the behavior of others from which the observer builds mental models. Constructivist Theory essentially states that students build new knowledge from current knowledge and students view new learning through the lens of previous experiences. In my teaching, I have designed classroom activities tailored toward both theories. Further, I believe it is critical for students to be able to link what is learned in the classroom to “real life” or to situations not traditionally discussed in relation to a particular subject, so that knowledge gained through the class is not compartmentalized. These are my current views on teaching, and as a teacher, I will seek to continually improve my teaching by engaging in Discipline-Based Education Research (DBER) and connecting with the educational research community. I am most interested in teaching courses on introduction to chemical engineering, numerical methods, and optimization, and am interested in developing courses on industrial microbial bioprocesses and design.