Lecture Summaries

1 Introduction Introduction to each other, the course objectives, logistics, and selection of a best meeting time. Also a brief introduction to systems biology, topics to be covered, and next week's topic.
2 Targeted Therapy in the Pre-Systems Biology Era For decades, cancer research focused on identifying oncogenes to promote a greater understanding of cancer, creation of prognostic biomarkers, and the advent of efficacious targeted therapies. Two successes have been the creation of the drug Herceptin for breast cancers containing an amplification of the oncogene HER2, and the drug Gleevec for Chronic Myeloid Leukemia (CML). This week we will cover two early papers that were instrumental in the creation of these therapies. In addition, we will discuss why these approaches have been less successful in the study of other diseases, and how systems biology can be used to help create therapies for complex diseases.
3 High-Throughput Data Acquisition I: Gene Sequencing and Functional Genomics A primary challenge in systems biology research is the collection of large amounts of quantitative data. The advent of new high-throughput gene sequencing technologies has made genomic information a reliable source of data for systems-level analyses. Today we will discuss two different approaches for gleaning biological insight from genomic data. The first paper uses a traditional approach: large-scale gene sequencing to identify genes that may be important in human gliomas. The second paper is an example of "functional genomics." These authors perform a genetic screen using short RNA hairpins to identify genes critical for breast cancer survival.
4 High-Throughput Data Acquisition II: Proteomics For many processes gene sequence or even gene expression does not provide insight into the activity or regulation of a cellular process. For example, post-translational modification is often used to alter protein activity or expression. These modifications can be dysregulated in disease states and are often the target of drug intervention. This week's topic is proteomic analysis. Both of the discussion papers use advanced mass spectrometry based techniques to quantify the activity, expression and modification states of large protein networks. The first paper analyzes the activity state of all human kinases to understand the efficacy of drugs used to treat CML, and the second paper uses a new technique, called CyTOF, to examine drug sensitivity across different hematopoietic lineages.
5 Gene Expression Analysis Gene expression analysis (typically performed by isolating and sequencing mRNAs) can provide insight into the functional "state" of a cell. This technique has been used for decades in cancer research for providing a molecular classification of tumors (i.e. subtyping). The first paper details the landmark discovery that gene expression profiling can be used to predict clinical outcome in breast cancer, and the second paper uses statistical analysis of many published gene expression profiles to find that almost any random gene signature can predict patient outcome. For both papers we will discuss the validity of the authors' claims and to what extent mechanistic insight can be gained from these data.
6 Network Biology Representation of complex systems as networks can be useful for visualizing their underlying structure, identifying functional units, and informing new hypotheses. Networks can be built from essentially any information: known protein-protein interactions, gene expression profiles, functional annotation, etc. The first paper constructs a network of human diseases and their underlying genetic properties, based on available Mendelian inheritance data. The second paper uses a combination of protein-protein interaction data and gene expression data to build networks used to predict metastasis of breast cancer.
7 Clustering Cluster analysis is essentially an organizational task aimed at identifying groups that are more similar to each other than to other groups. Using this information, we can then test the hypothesis that members of a group share similar properties. The first paper studies glioblastomas containing a variant of the Epidermal Growth Factor Receptor (EGFR) called "vIII." By clustering proteomic time-course data, these authors identify signaling responses found uniquely in EGFRvIII glioblastomas, and use this information to hypothesize novel therapies. The second paper uses clustering of "functional genetic signatures" (i.e. cellular responses to gene knockdown) to predict mechanisms of drug action for novel drug compounds and novel combination therapies.
8 Regression Modeling Regression analysis is commonly used to reduce data dimensionality and to identify complex relationships between independent and dependent variables. The first paper uses partial least squares regression to better understand the life/death decisions of liver cancer cells following cytokine treatment. The second paper uses a similar analysis to identify predictors of cell death in an in vivo model of inflammatory bowel disease.
9 Logic Modeling A current trend in the study of protein signaling pathways is to think of signaling networks as electrical circuits. Using this conceptual framework, many are using Boolean Logic (a form of algebra in which pathway activity is represented as "0" or "1") to understand signaling network topology and the logic of information flow through complex networks. The first paper uses classic Boolean Logic to identify new therapeutic targets in NF-kB signaling, and the second paper uses an extension of Boolean Logic called "Constrained Fuzzy Logic" (cFL) to understand pathway crosstalk downstream of complex cytokine cues. cFL is similar to Boolean Logic, but also allows for non-0 or 1 numbers.
10 Kinetic Modeling Modeling reaction kinetics requires detailed data and a refined understanding of the process of interest, and can provide mechanistic insight and clearly testable hypotheses. The first paper uses a kinetic model built using ordinary differential equations to understand the kinetics and "irreversibility" of mitotic onset. The second paper uses a similar type of model to determine how oscillations in p53 expression contribute to cell-fate decisions.
11 Structural Biology Structural biology is not typically included under the umbrella of systems biology, but these fields share similar characteristics. Furthermore, synthetic biology (next week's topic, which is typically grouped with systems biology) often requires detailed information about protein structure. The first paper is the latest in a series focused on the structure of EGFR, a common oncogene. Data in this paper shed light onto the mechanisms we discussed in Week 7 by which mutations in EGFR that are commonly seen in cancer function to activate the protein. The second paper is the recently solved crystal structure of a G protein coupled receptor (GPCR) bound to the full G protein heterotrimer. This structure leads the authors to propose a new (and controversial) mechanism of G protein activation and provides critical insight into the efficacy (or lack thereof) of many drugs.
12 Synthetic Biology The goal of synthetic biology is to create new and useful biological systems not found in nature. Today we will focus on two early reports from the field of synthetic protein engineering and biofuel production. In the first paper, the authors attempt to make synthetic protein-protein interactions to redirect signal transduction. The second paper uses genomic analyses to identify opportunities to circumvent toxicities faced when rewiring E. coli for biofuel production.
13 Field Trip Visit to Merrimack Pharmaceuticals We will visit Merrimack Pharmaceuticals, a bio-pharmaceutical company located in Kendall Square, Cambridge, MA. Merrimack has been a leader in the use of systems biology to inform drug development.
14 Systems Pharmacology and Network Medicine Systems Pharmacology, or Network Medicine, is the application of systems biology principles to drug design and rational targeted therapy. Both of today's papers use systems level analyses to rationally design new anti-cancer therapies. The first paper analyses all human kinases (i.e. the "kinome") to identify patterns of network rewiring following inhibition of MEK, one of a number of mitogen activated protein kinase kinases (i.e. MAPKK or MAP2K). This information is then used to design efficacious combinations for "triple-negative" breast cancers, those that do not express estrogen receptor, progesterone receptor, or the oncogene HER2. The second paper uses a large-scale genomic analysis of nearly 1000 human cancer cell lines to predict sensitivities to various clinically used anti-cancer drugs.
15 Final Class Students will present oral presentations, and we will review and discuss the course.