SBHD 2014 Program Book - Systems Biology of Human Disease 2014
Transcription
SBHD 2014 Program Book - Systems Biology of Human Disease 2014
INTERNATIONAL CONFERENCE ON JUNE 17–19 2014 sbhd2014.org SPONSORED BY: THE JOSEPH B. MARTIN CONFERENCE CENTER HARVARD MEDICAL SCHOOL 77 AVENUE LOUIS PASTEUR BOSTON, MA 02115 Contents Welcome . . . . . . . . . Schedule . . . . . . . . . Tuesday, June 17 . . Wednesday, June 18 Thursday, June 19 . Awards . . . . . . . . . . Attendee List . . . . . . . Speaker Abstracts . . . . Tuesday, June 17 . . Wednesday, June 18 Thursday, June 19 . Poster Abstracts . . . . . General Information . . . Floor Plans . . . . . HMS Campus Map . Area Street Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 10 10 12 14 16 20 36 36 45 54 62 184 184 185 186 2 International Conference on Systems Biology of Human Disease Welcome June 2014 Dear Colleagues, We are pleased to welcome you to the 2014 International Conference on Systems Biology of Human Disease (SBHD), a conference organized by the Council for Systems Biology in Boston (CSB2 ; www.csb2.org). This year’s three-day program includes platform talks, talks chosen from submitted abstracts, lightning talks, poster sessions and award presentations. We are particularly pleased by the large number of poster submissions and hope you spend some time to attend the poster sessions. The 2014 SBHD meeting is supported by the newly established Harvard Program in Therapeutic Sciences (HiTS). HiTS engages clinicians and scientists from multiple universities and Harvard affiliated hospitals in the application of systems biology approaches to therapeutic mechanisms and drug evaluation, with the ultimate goal of improving how existing treatments are personalized to individual patients and how drugs are developed for unmet medical needs. Achieving these goals will require deep understanding of cell signaling networks, disease processes and medicinal chemistry – the focus of the Laboratory of Systems Pharmacology – and fundamental improvements in the processes used to evaluate drugs via clinical trials – the focus of our Regulatory Sciences Program. As part of this latter effort we are sponsoring a special symposium “Challenges in Drug Evaluation and Regulatory Science: Applications to Multiple Sclerosis” on Thursday afternoon (June 19). We hope you will join us to hear how challenges arising from “adverse clinical events” were tackled by investigators close to the program that developed Alemtuzumab (Lemtrada) as a treatment for multiple sclerosis. This meeting would not be possible without the continuing alliance between HMS and its European partners, including the Helmholtz Alliance on Systems Biology (www.helmholtz.de/systemsbiology), DKFZ and BioQuant. In 2015 SBHD will once again be held in Heidelberg, after which it will return to Boston in 2016. Dates for future conferences will be announced during the proceedings, posted at www.csb2.org and shared by email (if you are a current or past SBHD attendee). We are grateful for Merrimack Pharmaceuticals’ continuing sponsorship of the CSB2 Prize in Systems Biology. This prize, which recognizes independent junior investigators who have exhibited exceptional promise, will be awarded in 2014 to Chris Bakal of the London Institute for Cancer Research. In addition, Chroma Technology Corp. has generously sponsored the first Anne Heidenthal Prize for Fluorescence Research awarded to Bernd Bodenmiller of the University of Zürich. Birgit Schoeberl from Merrimack and Georg Draude and Paul Millman from Chroma will present the awards prior to the awardees’ talks Boston, MA June 17-19, 2014 3 Welcome on June 18. As always, we welcome any ideas you might have for improving this conference. We are particularly aware that we have fewer social events than the Heidelberg-based versions of this meeting and we are trying to figure out how we can afford it in future years. Organizing an event such as SBHD requires an enormous amount of effort by a talented team, and we would like to thank Chris Bird, Laura Maliszewski, Jeremy Muhlich and Joleen Pugliese, as well as members of the Organizing Committee (see below). We hope that you have an enjoyable and exciting meeting, and look forward to seeing you in Germany in 2015. Yours sincerely, Peter Sorger Roland Eils SBHD 2014 Organizing Committee Peter Sorger, Harvard Medical School, Boston, USA Roland Eils, DKFZ, Heidelberg, Germany Bree Aldridge, Tufts, Cambridge, USA Leonidas Alexopoulos, National Technical University of Athens, Greece Philipp Bastiaens, MPI Dortmund, Germany Pascal Braun, TU München, Munich, Germany Suzanne Gaudet, Dana Farber Cancer Institute, Boston, USA Alexander Hoffmann, UC San Diego, San Diego, USA Ursula Klingmüller, DKFZ, Heidelberg, Germany Avi Ma’ayan, Mount Sinai Medical Center, New York, USA Uwe Sauer, ETH Zürich, Switzerland Birgit Schoeberl, Merrimack Pharmaceuticals, Cambridge, USA Luis Serrano, CRG Barcelona, Spain SBHD 2014 Local Organizing Team Chris Bird, Harvard Medical School, Boston, USA Laura Maliszewski, Harvard Medical School, Boston, USA Jeremy Muhlich, Harvard Medical School, Boston, USA Joleen Pugliese, Harvard Medical School, Boston, USA Funds to support this meeting were generously provided by the Harvard Program in Therapeutic Science (HiTS), DKFZ, BioQuant, Hemlholtz Association, and SystemsX.ch. Sup4 International Conference on Systems Biology of Human Disease Welcome port for travel costs for German students was provided by the German Federal Ministry of Education and Research (BMBF). We thank Birgit Schoeberl and Merrimack Pharmaceuticals, sponsor of the CSB2 Prize in Systems Biology and Paul Millman, Head of Chroma Technology Corp., sponsor of the Anne Heidenthal Prize for Fluorescence Research. Thanks also to Ulrike Conrad and Jan Eufinger for their assistance with planning and coordination of the meeting. Boston, MA June 17-19, 2014 5 Welcome Some important things to keep in mind: Speakers: • Please keep an eye on the podium timer and do not go over your allotted time. • Please bring your presentations or computers to the attention of AV staff at the front of the room well before the session begins. Everyone: • Please do not stand or sit in the aisles — there is sufficient seating for everyone. • Please use the microphone when asking questions of the speakers. • Public wireless access is available in all the conference rooms. Please speak to an AV staff member if you experience any difficulties. • In case of an emergency: – Please evacuate the room by following the nearest exit signs. – Upon exiting, head to the conference center assembly area (see map on facing page). – The assembly area is the MERCK Parking lot that is located to the left of the building as you exit from the main entrance. – Please report to the person wearing the green vest and holding “The Conference Center at Harvard Medical” sign. – Await further instructions. Drink tickets for the poster sessions: • Tickets in the back of your badge are for drinks during the poster session. You have two for each night. Non-alcoholic drinks do not require a ticket. You must be at least 21 years of age to order and/or consume drinks containing alcohol. 6 International Conference on Systems Biology of Human Disease Welcome Boston, MA June 17-19, 2014 7 Welcome 8 International Conference on Systems Biology of Human Disease Schedule Schedule Tuesday, June 17, 2014 Registration and continental breakfast 7:45 – 8:15 AM Welcome remarks 8:15 – 8:30 AM Roland Eils, German Cancer Research Center (DKFZ) Prenatal Environmental Exposure Induces Epigenetic Reprogramming with Consequences for Disease Risk Later in Children’s Life 8:30 – 9:10 AM John Albeck, University of California, Davis Linking the Dynamics of Kinase, Transcriptional, and Metabolic Networks in Single Cells 9:10 – 9:50 AM Franziska Michor, Dana-Farber Cancer Institute Evolution of the Cancer Genome 9:50 – 10:30 AM Coffee break 10:30 – 10:50 AM Dirk Drasdo, INRIA Paris / IZBI Leipzig 10:50 – 11:30 AM How Quantitative Modeling can Inform on Disease Pathogenesis: Lessons from Liver Leonidas Alexopoulos, National Technical University of Athens 11:30 – 12:10 PM Tackling Cartilage Degeneration using Systems Biology and Biomechanics Approach Lunch (bagged lunch provided) 12:10 – 1:00 PM Grégoire Altan-Bonnet, Memorial Sloan-Kettering Cancer Center Single-cell Analysis of the Dysregulation of Antigen Signaling in Chronic Lymphocytic Leukemia (CLL) 1:00 – 1:40 PM Olga Troyanskaya, Princeton University Cell-lineage and Tissue-specific View of Human Disease on the Whole-genome Scale 1:40 – 2:20 PM Kathryn Miller-Jensen, Yale University Analysis of Single-cell Secretion Reveals a Role for Paracrine Signaling in Coordinating Macrophage Response to TLR4 Stimulation (Selected speaker from poster abstracts – See poster #1) 2:20 – 3:00 PM 10 International Conference on Systems Biology of Human Disease Tuesday, June 17 Coffee break 3:00 – 3:20 PM Ariel Kniss, Georgia Tech Developing Microfluidic and Experimental Tools to Interrogate Intracellular T Cell Signaling using a Frequency Response Analysis Approach (Selected speaker from poster abstracts – See poster #3) 3:20 – 4:00 PM Lightning talks 4:00 – 4:40 PM Six one-slide, five-minute presentations by selected presenters from the Tuesday poster session: Loice Chingozha # 5 Jia-Ren Lin # 9 Jette Strasen # 13 Evan Daugharthy # 7 Dimitris Messinis # 11 Chun-Chao Wang # 15 Tuesday poster session and refreshments Adjourn Boston, MA 4:40 – 6:30 PM 6:30 PM June 17-19, 2014 11 Schedule Wednesday, June 18, 2014 Registration and continental breakfast 8:00 – 8:25 AM Welcome remarks 8:25 – 8:30 AM Christopher Sassetti, University of Massachusetts Medical School Systems Genetic Approaches to Understand Tuberculosis Pathogenesis 8:30 – 9:10 AM Douglas Lauffenburger, Massachusetts Institute of Technology Multi-Scale Systems Analysis of Cell-Cell Communication and Signaling in Complex Inflammatory Disease 9:10 – 9:50 AM Markus Covert, Stanford University High-sensitivity Measurements of Multiple Kinase Activities in Live single Cells 9:50 – 10:30 AM Coffee break 10:30 – 10:50 AM AJ Marian Walhout, University of Massachusetts Medical School Interspecies Systems Biology: Nutritional Networks 10:50 – 11:30 AM Bernd Bodenmiller, University of Zürich 11:30 – 12:10 PM Recipient of the Anne Heidenthal Prize for Fluorescence Research – sponsored by Chroma Technology Corp. Analysis of Single Cell Networks through Time and Space by Mass Cytometry Lunch (bagged lunch provided) 12:10 – 1:00 PM Kevin Janes, University of Virginia Linking Signal-transduction and Gene-expression Networks by Statistical Modeling 1:00 – 1:40 PM Chris Bakal, The Institute of Cancer Research, London Recipient of the CSB2 Prize in Systems Biology – sponsored by Merrimack Pharmaceuticals How Signaling Regulates Cell Shape, and how Cell Shape Regulates Signaling 1:40 – 2:20 PM Ursula Klingmüller, German Cancer Research Center (DKFZ) Unraveling the Risk of Epo in Lung Cancer 2:20 – 3:00 PM 12 International Conference on Systems Biology of Human Disease Wednesday, June 18 Coffee break 3:00 – 3:20 PM Sara Dempster, AstraZeneca Approaches for Integrative Analysis of Pharmacology and Genome-wide Molecular Profiling Data Applied to a Large Scale Screen of Cancer Cell Lines (Selected speaker from poster abstracts – See poster #2) 3:20 – 4:00 PM Lightning talks 4:00 – 4:40 PM Six one-slide, five-minute presentations by selected presenters from the Wednesday poster session: Jia-Yun Chen # 4 Yan Kou # 8 Miles Miller # 12 Kelly (Wen Li) Chen # 6 Carlos Lopez # 10 Lakshmi Venkatraman # 14 Wednesday poster session and refreshments Adjourn Boston, MA 4:40 – 6:30 PM 6:30 PM June 17-19, 2014 13 Schedule Thursday, June 19, 2014 Registration and continental breakfast 8:00 – 8:25 AM Welcome remarks 8:25 – 8:30 AM Mohammed AlQuraishi, Harvard Medical School From Genomes to Molecular Phenomes 8:30 – 9:10 AM Tim van Opijnen, Boston College Host-genetics and Fluid Bacterial Genomes Make Virulence and Drug-tolerance Context Dependent 9:10 – 9:50 AM Birgit Schoeberl, Merrimack Pharmaceuticals A Systems Biology Approach to Drug Development: Clinical Testing of Biomarkers for the anti-ErbB3 Antibody MM-121 9:50 – 10:30 AM Coffee break 10:30 – 10:50 AM Sabrina Spencer, Stanford University Single-cell Dynamics of the Proliferation-quiescence Decision 10:50 – 11:30 AM Luis Serrano Pubul, Center for Genomic Regulation, Barcelona 11:30 – 12:10 PM Structure-energy-based Predictions, Competition and Network Modelling of ERbMAPK Signalling, and its Implication in RASopathy and Cancer Missense Mutations Lunch (bagged lunch provided) Adjourn 14 12:10 – 1:00 PM 1:00 PM International Conference on Systems Biology of Human Disease Awards Awards The CSB2 Prize in Systems Biology sponsored by Merrimack Pharmaceuticals This prize is awarded to Chris Bakal, Ph.D., Leader of the Dynamical Cell Systems Team within the Division of Cancer Biology at the The Institute of Cancer Research. Chris Bakal was chosen for his contributions to the development and implementation of new statistical methods to investigate signals regulating cellular behaviors by analyzing features of cells captured by imaging. In his earlier work, he showed how single-cell measurements of JNK activity under a multitude of genetic perturbations allowed the inference of a JNK signaling network. In the past year, the power of these new statistical methods was highlighted in a publication by his group in Nature Cell Biology. The statistical methods they developed enabled a quantitative description of cell shape complexity, showing that cells only take on a small set of discrete shapes, as well as identification of key regulators of cell shape heterogeneity. In all of his work, Chris Bakal combines the power of single-cell observations by microscopy with that of statistics to gain new insights in signal transduction and cell biology. The Anne Heidenthal Prize for Fluorescence Research sponsored by Chroma Technology Corp. Professor Bernd Bodenmiller is the first awardee of the “Anne Heidenthal Prize,” which is being generously sponsored by Chroma Technology Corp. The prize is awarded annually in memory of Anne Neumeyr-Heidenthal, a molecular and cell biologist at Chroma Technology Corp. The Anne Heidenthal Prize recognizes promising young scientisits who are advancing the field of quantitative biology through the use of fluorescence and advanced imaging technologies. Bernd Bodenmiller is the SNSF Assistant Professor for Quantitative Biology in the Institute of Molecular Life Sciences at the University of Zürich. As a postdoctoral fellow with Gary Nolan, Bernd was instrumental in the development of mass cytometry as means to perform high multiplicity profiling of human immune cells. In his own lab, Bernd has led the development of novel methods for imaging cells and tissues using mass cytometry, a revolutionary approach to single-cell analysis. 16 International Conference on Systems Biology of Human Disease Previous Awardees 2013 James Locke, Cambridge University, Cambridge UK 2012 Hana El-Samad, UCSF University of California, San Francisco 2011 Hang Lu, Georgia Institute of Technology 2010 Marcus Covert, Stanford University Melissa Kemp, Georgia Tech & Emory University 2009 Ernest Fraenkel, Massachusetts Institute of Technology Nathanael Gray, Dana Farber Cancer Institute 2008 Gavin MacBeath, Harvard University Aneil Mallavarapu, Harvard Medical School Boston, MA June 17-19, 2014 17 Awards 18 International Conference on Systems Biology of Human Disease Speaker Abstracts Speaker Abstracts Prenatal Environmental Exposure Induces Epigenetic Reprogramming with Consequences for Disease Risk Later in Children’s Life Roland Eils German Cancer Research Center (DKFZ) Environmental factors can cause persistent perturbations of regulatory pathways with the consequence of altered disease susceptibility. Prenatal exposure to environmental factors may increase the disease risk later in children’s life. It is currently discussed that the transient effect of environmental exposure early in life can be preserved by epigenetic mechanisms that may silence or activated disease relevant regulatory pathways in a persistent way. Here, we studied genome-wide, environmentally induced epigenetic changes and their functional relevance for disease risks later in life within a longitudinal mother-child birth cohort. Among others, we studied gene environment interactions induced by maternal smoking during pregnancy. By integrated analysis of longitudinal whole genome bisulfite sequencing, ChIP-sequencing and RNA sequencing we unraveled a genome-wide epigenetic stable reprogramming converging upon transcriptional enhancers. We found a number of disease related pathways deregulated, among them Wnt signaling, which is involved in the airway inflammatory response to cigarette smoke in smoking mothers as well as their newborn children. An association between epigenetic reprogramming of genes within the Wnt signaling pathway already at time of birth and the development of impaired lung function later in children’s life can be shown. 36 International Conference on Systems Biology of Human Disease Tuesday, June 17 Linking the Dynamics of Kinase, Transcriptional, and Metabolic Networks in Single Cells John Albeck University of California, Davis As single-cell technologies expand, it is becoming clear that many cellular signaling events are very dynamic, necessitating a time-lapse approach. I will present our work on the single-cell kinetics of two kinases - ERK and AMPK - that play key roles in the response to targeted cancer therapies aimed at disrupting cellular growth, proliferation, and homeostasis. Induced by growth factor stimulation, ERK activity is a central controller of transcription factors involved in oncogenesis, including Myc, Fra-1, and Egr-1. We show that at the single-cell level, each of these factors interprets ERK dynamics differently, leading to a diversity of cellular states within a genetically homogeneous population. ERK pathway inhibitors, now being evaluated for use in multiple cancers, modulate ERK dynamics differentially and redefine the repertoire of cellular states in unique ways. AMPK responds to cellular energy deprivation, and we show that direct inhibition of glycolysis results in a strikingly regular modulation of AMPK activity and metabolic state. In contrast, PI3K and mTOR inhibitors, another key class of targeted therapies, lead to highly disordered disruption of metabolic dynamics. Together these findings underscore the concept that, despite the chemically specific of modern targeted cancer therapies, their usefulness may be limited by the highly variable kinetics that they induce within cellular populations, resulting in sub-optimal, heterogeneous responses. Boston, MA June 17-19, 2014 37 Speaker Abstracts Evolution of the Cancer Genome Franziska Michor Dana-Farber Cancer Institute 38 International Conference on Systems Biology of Human Disease Tuesday, June 17 How Quantitative Modeling can Inform on Disease Pathogenesis: Lessons from Liver Dirk Drasdo INRIA Paris / IZBI Leipzig Systems biology has opened up new ways of understanding disease processes based on close iterations between experimentation and mathematical modelling. So far, its focus has mainly been on molecular processes. The combination of modern imaging modalities with image processing and analysis (Hammad et. al. Arch. Toxicol. 2014; Hoehme and Drasdo, Bioinformatics 2010), and mathematical models opens up a promising new approach towards a quantitative understanding of pathologies and of disease processes that includes the multicellular tissue level. For illustration we will consider three examples of interdisciplinary approaches integrating biological models and mechanisms of processes contributing to disease progression at various scales within mathematical modelling frameworks. In the first example a multi-cellular spatial temporal model predicts within a systems biology approach a previously not recognized and subsequently validated order principle underlying liver regeneration after drug-induced damage, as it occurs for example after overdosing acetaminophen (paracetamol) (Hoehme et. al., PNAS, 2010). The second example will present a mathematical model integrating information from the spatial temporal model of the first example with the chemical reactions known to detoxify liver from ammonia in health liver during the destruction and subsequent regeneration process, that succeeded to indicate the lack of an important reaction (Schliess et. al., Hepatology, 2014). Experiments triggered by this model prediction led to finding of a so far unrecognized good candidate reaction that might be clinically utilized in case of hyperammonemia. The final example will address the spatial-temporal molecular control of the regeneration process within a mechanistic multi-scale model spanning the molecular, cellular, tissue and body scale. The tissue model involved in each of these examples represents each cell individually as biophysical entities and is hence able to integrate equally physical and biological information. The example demonstrates that multi-cellular models are so far able to falsify hypotheses and guide towards the most informative experimental design. Boston, MA June 17-19, 2014 39 Speaker Abstracts Tackling Cartilage Degeneration using Systems Biology and Biomechanics Approach Leonidas Alexopoulos, Stavroula Samara, Ioannis N Melas, Elisavet Chatzopoulou, Theodore Sakalleropoulos, Alexander Mitsos, Zoe Dailiana, Panagoula Kollia National Technical University of Athens Arthritis is the leading cause of disability affecting one of every six people. The most common type of arthritis is osteoarthritis (OA), a painful disease, in which cartilage loses its mechanical integrity. It is evident that OA is not just a natural cause of aging but an imbalance of signaling mechanisms that leads to cartilage degeneration. Two different approaches are proposed to study OA: Systems Biology and Biomechanics. The systems biology approach uses high throughput proteomics and an optimization formulation in order to: i) screen compounds and identify major catabolic players for cartilage and ii) identify signaling mechanisms responsible for cartilage degeneration. The biomechanics approach is used to quantify the mechanical properties of cartilage on a custom multi indentation device. The combination of those approaches links a tissue phenotype (degeneration) with the signaling network and sheds a light into possible treatments of OA. Methods: Systems Biology: Chondrocytes are isolated and treated with a diverse set of 90 catabolic and anabolic stimuli. 18 phopshoproteins and 60 cytokines are measured using multiplex technology (Luminex) with custom multiplex assays (ProtATonce). Data are normalized and analyzed using PLSR, PCA, clustering algorithm, and integer linear programming formulation. Signaling pathways were constructed. Biomechanics: A multiindentation device than handles up to 24 cartilage disk explants is built and an analytical, linear elastic model and a biphasic model is used to quantify Young’s modulus, Poisson’s ratio, and hydraulic permeability. Results: The biomechanics approach quantifies the mechanical degeneration of the tissue whereas the systems biology approach identifies pathways responsible for cartilage degeneration such as the known IL1a, IL1b, and TNFa but also the less known TLR stimuli. Closer look into the pathways constructed by the ILP formulation identify possible ways of reversing the loss of mechanical properties via targeting of the NFkB and possibly the p38 pathway. Conclusions: The combination of Systems Biology and Biomechanics was able to identify new players and related pathways to cartilage degeneration. Our approach sheds a light into possible treatments of osteoarthritis and suggests novel therapeutic interventions of a disease with strong medical, economic, and social impact. 40 International Conference on Systems Biology of Human Disease Tuesday, June 17 Single-cell Analysis of the Dysregulation of Antigen Signaling in Chronic Lymphocytic Leukemia (CLL) Carly G. Ziegler, M. Lia Palomba & Grégoire Altan-Bonnet Memorial Sloan-Kettering Cancer Center Chronic Lymphocytic Leukemia (CLL) is characterized by a perturbed B-cell receptor mediated signaling machinery, driving B cell accumulation in the blood and lymphoid organs. We applied phosphatase inhibition to activate and probe ex vivo the signaling response of human primary B cells with single cell resolution. We found that B cells (from CLL patients or from healthy donors) rapidly present a bimodal response in proximal kinase phosphorylation. Using drug inhibitors, we validated the existence of a proximal positive feedback in kinase activation, whose existence is supported by the observed hysteresis in BCR signaling. We then documented systematically the kinetics of phosphorylation/dephosphorylation within the BCR pathway to build and constrain a universal model of BCR signaling. Our model predicted that all B cells (either from healthy donors or CLL patients) respond equivalently, once normalized for varied degrees of receptor clustering, as quantified by super-resolution microscopy. Our study highlights the role of tonic signaling in CLL B cells, and a potential defect in negative selection as a major driving event for oncogenesis in CLL. Boston, MA June 17-19, 2014 41 Speaker Abstracts Cell-lineage and Tissue-specific View of Human Disease on the Whole-genome Scale Olga Troyanskaya Princeton University An immense molecular complexity forms the foundation of human disease. Our goal is to interpret and distill this complexity through accurate modeling of molecular networks and pathways, particularly those in which malfunction promotes the emergence of complex human disease. Although cell-lineage-specific gene expression and function underlie the development, function, and maintenance of diverse cell types within an organism, highthroughput data are rarely resolved with respect to specific cell lineages. In this talk, I will focus on our recent work developing integrative approaches that leverage functional genomics data collections to study how cellular pathways function in diverse cell types. Our work accounts for cell-lineage specificity in integrated models of gene expression and protein networks, including predictions of tissue-specific and cell-lineage-specific gene expression in human with accuracies higher than those of high-throughput experimental studies. I will also discuss how integrated analysis of functional genomics data can be leveraged to study tissue-lineage-specific protein function and interactions and to identify genes involved in disease in a way complementary to quantitative genetics approaches. 42 International Conference on Systems Biology of Human Disease Tuesday, June 17 Analysis of Single-cell Secretion Reveals a Role for Paracrine Signaling in Coordinating Macrophage Response to TLR4 Stimulation Qiong Xue, Yao Lu, Markus Eisele, Nafeesa Khan, Rong Fan, Kathryn Miller-Jensen Yale University Cell populations produce reliable biological responses despite significant levels of cellto-cell heterogeneity. These biological responses often involve intermediate extracellular signals in which cells secrete and respond to the same factor. The propagation of intermediate signals by extracellular signaling has a significant impact on the collective cellpopulation response, but the contribution of autocrine versus paracrine signaling remains difficult to analyze. Here we combined multiplexed, nanowell single-cell secretion measurements with cell-population data to distinguish the role of paracrine signaling in shaping the inflammatory cytokine secretion profile in a monocytic cell line (U937) and primary human monocyte-derived macrophages (MDMs) in response to toll-like receptor 4 (TLR4) stimulation with lipopolysaccharide (LPS). Loss of paracrine signaling upon single-cell isolation in nanowells significantly reduced secretion of a subset of LPS-stimulated cytokines, including interleukin-6 (IL-6) and IL-10, in both U937 and primary human MDM cells. Using graphical Gaussian modeling of LPS-stimulated single-cell data, we identified tumor necrosis factor-α (TNF-α) and IL-1β as central nodes in the LPS-stimulated paracrinesignaling network, which was further validated experimentally. In both U937 and MDM single cells, a small fraction of cells accounted for the majority of the total TNF-α output in the cell population, but paracrine signaling from this small subset amplified secretion and reduced variability in the downstream dependent signals. Our results demonstrate that a small subpopulation of high-responding cells appears to drive the innate immune response in macrophages through paracrine signaling. Thus, paracrine signaling results in a coordinated population response that is absent in isolated single cells, underscoring the importance of inter-cellular signaling in determining cellular behaviors. Boston, MA June 17-19, 2014 43 Speaker Abstracts Developing Microfluidic and Experimental Tools to Interrogate Intracellular T Cell Signaling using a Frequency Response Analysis Approach Ariel Kniss, Loice Chingozha, Hang Lu, Melissa Kemp Georgia Institute of Technology Introduction: T cells are part of the adaptive immune response that recognize and fight pathogens in the body. T cell activation is a complex process resulting in the induction of a proliferative response, encoded by the dynamics of multiple signaling cascades and reliant on second messenger molecules such as calcium and hydrogen peroxide. Frequency response analysis (FRA), developed in control engineering, has been previously allowed for inference of cellular signaling network structure but has been experimentally difficult to apply to suspension T cells. We have designed a FRA experimental and computational platform for future use in determining the dominant feedback controls present in T cell signaling. Materials and Methods: To develop the experimental precision required for this approach, we use a microfluidic device capable of applying an oscillatory chemical cue as an input signal while enabling visualization of T cells through time. In this work we apply H2 O2 at a range of driving frequencies to Jurkat T cells while monitoring cytoplasmic calcium concentration. We then perform spectral analysis of the single cell calcium traces to determine the output signal response of the system. Results: Calcium signaling is found to respond differently to 100µM H2 O2 at specific driving frequencies of 16.7mHz (1 min period), 8.3mHz (2 min period), and 2.8mHz (6 min period). With the single cell spectral analysis, subpopulations of cells emerge in response to the same driving frequency and between different experiments. Cells are entrained to frequencies below 8.3mHz but attenuate H2 O2 signals above 16.7mHz. Conclusion: With the use of microfluidic technology and initial spectral analysis, we have determined that T cells have a cutoff frequency between 1 and 2 minutes when interrogated with oscillatory H2 O2 stimulus. This suggests downstream transcriptional effects are dependent on the frequency of environmental cues, with high frequency signals filtered out as potential noise. 44 International Conference on Systems Biology of Human Disease Wednesday, June 18 Systems Genetic Approaches to Understand Tuberculosis Pathogenesis Christopher Sassetti University of Massachusetts Medical School Mycobacterium tuberculosis is an obligate human pathogen that infects a significant fraction of the world’s population. Throughout its coevolution with humans, both host and pathogen have imposed direct selective pressures on each other. As a result, the competition between M. tuberculosis and the immune system can be largely viewed as an interaction between two genomes. Based on this concept, we are working to understand the mechanisms underlying tuberculosis disease progression through the identification of intra- and inter-species genetic interaction (GI) networks. Bacterial GIs can be mapped at a genome-wide scale using high-density transposon mutagenesis followed by deep sequencing. Networks defined during infection reflect the physiology of the bacterium in this specific environment and can be used to delineate condition-specific pathway structure. We have used this to understand complex adaptations to the host, such as carbon acquisition and redox homeostasis. By subjecting bacterial libraries to selection in mice with systematically-varied genomes, inter-species GIs can also be identified, which associate bacterial adaptations to the immune functions that impose the corresponding pressure. We are exploring this approach as a new strategy to discover and functionally characterize host susceptibility loci. Boston, MA June 17-19, 2014 45 Speaker Abstracts Multi-Scale Systems Analysis of Cell-Cell Communication and Signaling in Complex Inflammatory Disease Douglas Lauffenburger Massachusetts Institute of Technology Complex inflammatory diseases such as arthritis, diabetes, and endometriosis involve interactions of immune system cells with tissue cells via molecular communication processes, and resulting disease consequently arises from dysregulated signaling network activities governing pathological cell behaviors. We are employing combined experimental/computational systems biology approaches to understand these communication and signaling processes at multiple scales, integrating across multiple intracellular pathways, multiple extracellular factors, and multiple cell types. 46 International Conference on Systems Biology of Human Disease Wednesday, June 18 High-sensitivity Measurements of Multiple Kinase Activities in Live single Cells Markus Covert Stanford University Increasing evidence has shown that population dynamics are qualitatively different from single cell behaviors. Reporters to probe dynamic, single cell behaviors are desirable, yet relatively scarce. Here we describe an easy-to-implement and generalizable technology to generate reporters of kinase activity for individual cells. Our technology converts phosphorylation into a nucleocytoplasmic shuttling event that can be measured by epifluorescence microscopy. Our reporters reproduce kinase activity for multiple types of kinases, and allow for calculation of active kinase concentrations via a mathematical model. Using this technology, we made several experimental observations that had previously been technically unfeasible, including stimulus-dependent patterns of c-Jun N-Terminal Kinase (JNK) and Nuclear Factor kappa B (NF-κB) activation. We also measured JNK, p38 and ERK activities simultaneously, finding that p38 regulates the peak number, but not the intensity, of ERK fluctuations. Our approach opens the possibility of analyzing a wide range of kinase-mediated processes in individual cells. Boston, MA June 17-19, 2014 47 Speaker Abstracts Interspecies Systems Biology: Nutritional Networks AJ Marian Walhout University of Massachusetts Medical School How does our diet affect our physiology? And what are the networks involved in interpreting nutritional state and executing systems level phenotypic outputs? We have discovered that physiology of nematode C. elegans is dramatically different, depending on which bacterial diet it consumes. We pioneered a novel ‘interspecies systems biology’ approach by which we started to delineate the networks involved in both the worm and the bacteria. These networks are directly relevant not only to human inborn errors of metabolism, but may also illuminate the biology of complex diseases such as type 2 diabetes and obesity. 48 International Conference on Systems Biology of Human Disease Wednesday, June 18 Analysis of Single Cell Networks through Time and Space by Mass Cytometry Bernd Bodenmiller University of Zürich Recipient of the Anne Heidenthal Prize for Fluorescence Research – sponsored by Chroma Technology Corp. Tumors are heterogeneous assemblies of multiple cell types that interact and communicate with each other to achieve disease states. Many aspects of tumor development to metastasis depend on these cell interactions in unique microenvironments. Especially signaling networks, forming the core of cellular decision making, are controlled by cell-to-cell communication. To study and understand the decision making in the microenvironments, single cell analysis technologies are needed that allow to measure cell type, signaling network state and other cellular processes with spatial resolution. Mass cytometry is a recent single cell mass spectrometry approach that enables to measure up to 100 proteins and their modifications simultaneously using isotopically pure rare earth metals as reporters on antibodies. Previously, only cells in suspension could be analyzed by mass cytometry, and thus essential information on cell location and cell-to-cell interactions was lost. We have now coupled immunocytochemical and immunohistochemical methods with high-resolution laser ablation to mass cytometry. The approach now enables the simultaneous imaging of up to 100 proteins and phosphorylation sites at a sub-cellular resolution. We use mass cytometry to study the signaling networks activated during the epithelial-mesenchymal transition (EMT), a process driving the formation of metastasis, in model systems and within their native microenvironment in human breast cancer tumors. Imaging mass cytometry will enable the analysis into how cellular assemblies generate phenotypes in health and disease and will support the transition of medicine towards individualized molecularly targeted therapies. Boston, MA June 17-19, 2014 49 Speaker Abstracts Linking Signal-transduction and Gene-expression Networks by Statistical Modeling Kevin Janes University of Virginia Changes in cell state are initiated by environmental stimuli and executed by a combination of signaling and regulated gene expression. In this talk, we propose a data-driven approach to examine the interface between the effector proteins of cell signaling and the transcription factors that trigger gene expression. We built a tri-linear statistical model that connects time-varying, multiparameter measurements of signaling with dynamic expression clusters obtained by microarray profiling. The model allowed us to link signaling and transcriptional pathways that were jointly regulated by crosstalk from proinflammatory cytokines and growth factors. We will discuss ongoing experiments that seek to test the hypothesis that – for one such transcriptional cluster that is stimulated by tumor necrosis factor (TNF) and antagonized by insulin – crosstalk occurs through novel GSK3-mediated phosphorylation of the endodermal transcription factor GATA6. 50 International Conference on Systems Biology of Human Disease Wednesday, June 18 How Signaling Regulates Cell Shape, and how Cell Shape Regulates Signaling Chris Bakal The Institute of Cancer Research, London Recipient of the CSB2 Prize in Systems Biology – sponsored by Merrimack Pharmaceuticals The primary goal of the Bakal laboratory is to understand the generation of cell form during migration. Our laboratory’s philosophy is that a comprehensive understanding of cell migration in both health and disease can only occur if we ascertain the architecture and dynamics of signaling networks that regulate, and spatiotemporally co-ordinate, the specific morphogenetic processes underpinning cell migration. Towards the aim of mapping the signaling networks that regulate cell shape, and how their dynamical behavior results in cell migration, we utilize two approaches. Firstly, we generate datasets that describe single cell shape following systematic gene perturbation using RNA interference (RNAi). We have shown that datasets describing cell shape following gene depletion can be used to: i) rapidly characterize how individual genes contribute to the cell shape changes required for migration; ii) map networks of functional interactions between these genes. Secondly, we perform high-throughput screens where we quantify single cell shape and signaling activity across diverse cell lines in the absence of perturbation, and exploit naturally occurring phenotypic heterogeneity to describe quantitative and predictive relationships between cell morphogenesis and signaling dynamics. Importantly, we believe that rigorous computational analysis of datasets describing cell shape will generate new systems-levels insights into cell shape and motility beyond the role of individual genes or the description of signaling networks. Here I will discuss how single cell imaging experiments and statistical analyses have revealed a novel role for morphological heterogeneity in the regulation of the inflammatory response in breast cancer cells, and show how morphological heterogeneity is an evolvable process regulated by signaling networks. Boston, MA June 17-19, 2014 51 Speaker Abstracts Unraveling the Risk of Epo in Lung Cancer Ursula Klingmüller German Cancer Research Center (DKFZ) The hormone erythropoietin (Epo) has been widely used for the treatment of anemia. However, in the context of chemotherapy related anemia its safety is controversially discussed. We establish a detailed dynamic pathway model of Epo-mediated JAK-STAT signaling in primary erythroid progenitor cells and adapt this to Epo-induced JAK-STAT signaling in lung cancer cells. Systematic analysis and comparison of model parameters reveals that negative feedback regulation is severely altered in the cancer cells providing a possibility to selectively perturb Epo-induced signaling in lung cancer. Furthermore, we utilize our dynamic pathway model of Epo-EpoR interaction to identify erythropoiesis stimulating agents (ESA) that due to their specific binding properties might provide a safer treatment option in lung cancer. In summary, our approach demonstrates that through rigorously quantitative dynamic pathway models we can contribute to design improved treatment options for patients. 52 International Conference on Systems Biology of Human Disease Wednesday, June 18 Approaches for Integrative Analysis of Pharmacology and Genome-wide Molecular Profiling Data Applied to a Large Scale Screen of Cancer Cell Lines Sara Dempster, Jonathan Dry AstraZeneca AstraZeneca has partnered with the Wellcome Trust Sanger Institute to screen the AstraZeneca oncology portfolio across a diverse panel of 1000 cancer cell lines. The cell lines have been profiled by gene expression arrays, whole exome sequencing, and array CGH. The dataset provides a unique opportunity to identify novel molecular markers of drug sensitivity and resistance in pre-clinical models. We are integrating the datasets to identify novel molecular markers to inform hypotheses for patient segmentation. For example, one goal is to identify molecular features that are associated with differences in specificity and efficacy of AZ compounds that target the same pathways. Here, we focus on the application of methods that will support these goals. Because there are hundreds of thousands of molecular features, we need to reduce the dimensionality by grouping features likely to have redundant functional impact in order to improve the power of our analysis. To address this challenge, we are exploring approaches that leverage known biological pathways and networks such as HotNet and NBS as well as unsupervised clustering approaches such as iCluster Plus. We will present different approaches for applying the HotNet algorithm to the data, show initial results, and propose next steps. Boston, MA June 17-19, 2014 53 Speaker Abstracts From Genomes to Molecular Phenomes Mohammed AlQuraishi Harvard Medical School Large-scale genomic and proteomic initiatives have respectively generated thousands of genome sequences and millions of protein-protein binding measurements, yet it remains difficult to determine the effects of mutations on protein networks. We have developed an analytical framework that integrates and reconciles these distinct data modalities to model the human SH2-phosphoprotein network in normal and cancer cells, and experimentally validated novel predicted interactions and the effects of cancer mutations on binding affinity. Our analysis indicates that cancer mutations perturb SH2 networks in a bimodal fashion depending on tissue type, surgically rewiring individual interactions to disrupt connected sub-networks in one mode, and bluntly disrupting the function of one signaling protein in the other mode. The effects of cancer mutations also vary by protein type, frequently resulting in gain of new interactions in phosphoproteins, and almost exclusively loss of interactions in SH2 domains. Our analytical framework represents a new approach to the interpretation of genomic data that synthesizes genetics and biophysics. 54 International Conference on Systems Biology of Human Disease Thursday, June 19 Host-genetics and Fluid Bacterial Genomes Make Virulence and Drug-tolerance Context Dependent Tim van Opijnen Boston College In the last decade the increased availability of complete bacterial genomes has demonstrated a distinction between a species’ core-genome (the pool of genes that is shared by all members of a species) and its pan-genome (the global gene repertoire of a species). This implies that, even within species, genomes are fluid and new elements are easily integrated into the genomic network. This plasticity greatly expands the possible geneticinteraction space and consequential phenotypes. To reduce such complexity the environment and the genetic background are often experimentally controlled, which ignores that in natural environments are rarely constant, that genomic content is subject to change and that in effect genotype-phenotype relationships are highly context dependent. Here we focus on the bacterial pathogen Streptococcus pneumoniae and show the importance of the host-environment as well as bacterial genomic content for survival, disease induction and drug tolerance. S. pneumoniae is a human nasopharyngeal commensal and respiratory pathogen. It triggers pneumococcal pneumonia, meningitis, and septicemia, resulting in around 1 million deaths annually among children under 5 years of age, and another approximately 0.5 million deaths among the immunocompromised and the elderly. This enormous disease burden places S. pneumoniae among the most important bacterial pathogens worldwide. By exploring different pneumococcal strains with genome-wide transposon mutagenesis coupled to massive parallel sequencing (Tn-seq), expression experiments, genetic interaction mapping and droplet-based microfluidics, we show that the genetic background drastically alters genotype-phenotype relationships – including essentiality of genes – and has a large impact on the way the genome functions, affecting diverse processes such as metabolism, drug tolerance and virulence. Besides genomic content, host genetics confounds bacterial virulence. Children with sickle cell disease (SCD) have an extremely high risk of fatal pneumococcal infection, yet the reason for this increased risk is largely unknown. By performing whole genome sequencing on over 300 pneumococcal strains isolated from SCD patients over the last 20 years, we determined that gene content has changed and that a shift towards non-vaccine serotypes has taken place while invasive capacity has been retained. Additionally, a murine SCD model coupled to Tn-seq identified specific streptococcal genes that directly Boston, MA June 17-19, 2014 55 Speaker Abstracts link immune system deficiencies and different availability of micronutrients in the SCD host, to an increase in bacterial virulence. Moreover, vaccination experiments demonstrate that the SCD host may be unable to raise a sufficient immune response due to loss of potential antigens that are dispensable for infection in high-risk individuals. These studies highlight the importance of identifying context-dependent phenotypes on a broader scale. Applying our functional genomics framework will not only provide to help unravel fundamental principles of genetic networks resulting from natural variation but for bacteria it means we will be able to better predict virulence, and drug-tolerance of individual strains on a species wide-level. 56 International Conference on Systems Biology of Human Disease Thursday, June 19 A Systems Biology Approach to Drug Development: Clinical Testing of Biomarkers for the anti-ErbB3 Antibody MM-121 Birgit Schoeberl Merrimack Pharmaceuticals Using a Systems Biology approach we identified ErbB3 as the most critical activator of phosphoinositide 3-kinase (PI3K) signaling within the ErbB signaling network. Based on this insight Merrimack Pharmaceuticals designed a monoclonal antibody, MM-121, which blocks HRG (heregulin) and BTC (betacellulin) induced signaling. (Schoeberl et al., 2009) In order to understand which patients would respond to MM-121, we identified a set of five preclinically defined biomarkers (EGFR, ErbB2, ErbB3, HRG and BTC) that allowed us to distinguish between responding vs. non-responding cancer cell lines grown as xenografts. Recently, MM-121 completed three Phase 2 clinical trials to determine if patients with advanced malignancies would derive benefit from the addition of MM-121 to their standard therapy. Here we will describe the result of translating the preclinical biomarker findings into the clinical development of MM-121. Among 464 patients who participated in our ovarian, breast and lung clinical trials we obtained HRG and BTC measurements by RNA-ISH and RT-PCR and quantitative receptor expression by qIHC of ErbB3, EGFR and HER2 in a large fraction of the patients. Of the five biomarkers, the most predictive of response was HRG mRNA: patients with detectable HRG in pre-treatment biopsies or high HRG in archived tissue blocks responded poorly to standard-of-care therapy and benefited most from MM-121. In addition, benefit was largely restricted to patients with low ErbB2 (<200,000 receptors per cell) which we had predicted preclinically. In fact, the insight that the MM-121 activity is limited by high HER2 levels was the reason for Merrimack to design MM-111, a ultra-potent ErbB3 inhibitor in HER2 amplified tumors. In summary, we found that across the different Phase 2 trials heregulin seems to be a biomarker for poor response to standard of care therapy in HER2 low tumors and a potential predictor of clinical benefit from MM-121 in late-stage ovarian, lung, and breast cancers which is in accordance to our preclinical studies. To our knowledge, this is the first example of a drug designed and clinically tested based on Systems Biology insights. 1. Schoeberl, B., Pace, E. a, Fitzgerald, J. B., Harms, B. D., Xu, L., Nie, L., Nielsen, U. B. (2009). Therapeutically targeting ErbB3: a key node in ligand-induced activation of the ErbB receptor-PI3K axis. Science Signaling, 2(77), ra31. doi:10.1126/scisignal.2000352 Boston, MA June 17-19, 2014 57 Speaker Abstracts Single-cell Dynamics of the Proliferation-quiescence Decision Sabrina Spencer Stanford University Tissue homeostasis in metazoans is regulated by transitions of cells between quiescence and proliferation. The hallmark of proliferating populations is progression through the cell cycle, which is driven by Cyclin-dependent kinase (CDK) activity. I will discuss our recent development of a live-cell sensor for CDK2 activity and the finding that proliferating cells bifurcate into two populations as they exit mitosis. Some cells immediately commit to the next cell cycle by building up CDK2 activity from an intermediate level, while other cells lack CDK2 activity and enter a transient state of quiescence. This bifurcation is directly controlled by the CDK inhibitor p21 and is regulated by mitogens during a restriction window at the end of the previous cell cycle. Thus, cells decide at the end of mitosis to either start the next cell cycle by immediately building up CDK2 activity or to enter a transient G0-like state by suppressing CDK2 activity. 58 International Conference on Systems Biology of Human Disease Thursday, June 19 Structure-energy-based Predictions, Competition and Network Modelling of ERbMAPK Signalling, and its Implication in RASopathy and Cancer Missense Mutations Luis Serrano Pubul Center for Genomic Regulation, Barcelona Different cell types share many of their signaling molecules yet can respond differently to the same stimuli, through ways not fully understood. One hypothesis is that differences in protein concentrations could change signaling outputs if there is competition at a critical network branching point. Here we analysed the extent of competing interactions in the ErbB signaling network. We extended the latest high-confidence network, identified domains and linear sequence motifs mediating binary interactions, and their affinities. Using three-dimensional structures the network was dissected into compatible and mutually exclusive interactions. We experimentally determined protein abundances for a large fraction of the 198 network nodes in three cell types, and uncovered a high degree of competition. In a combined experimental and computational network modelling approach we showed for one competing node that protein level perturbations remodeled downstream signaling flows. Our results suggest that protein level variation at competing nodes contributes to cell type-specific signaling responses. Using the 3D reconstructed network we investigated a group of developmental disorders caused by by germline mutations in 15 genes encoding proteins of the Ras/mitogen activated protein kinase (MAPK) pathway frequently involved in cancer. Using the protein design algorithm FoldX, we predict that most of the missense mutations with destabilizing energies are in structural regions that control the activation of proteins, and only a few are predicted to compromise protein folding. We find a trend in which energy changes are higher for cancer compared to RASopathy mutations. In summary, we suggest that quantitative rather than qualitative network differences determine the phenotypic outcome of RASopathy compared to cancer mutations. Our work shows the power of combining network analysis, with structural information and protein design in dissecting cell signaling and in disease. Boston, MA June 17-19, 2014 59 Speaker Abstracts 60 International Conference on Systems Biology of Human Disease Poster Abstracts Poster Abstracts Alphabetical list of poster presenters Presenter Orfeas S. Aidonopoulos Ozan Alkan Matthew Anderson Vinayagam Arunachalam Gnanaprakash Balasubramanian Joerg Bartel François Bertaux Benjamin Boucher Madison Brandon Leslie Brick John Burke Alexzandrea Buscarello Jaime Campos Natalie Catlett Jia-Yun Chen Kelly W. Chen Loice Chingozha Michael Chou Giovanni Ciriello Pau Creixell Kristina D’Agostino Evan Daugharthy Sara Dempster Gautam Dey Madeline Diekmann Sulav Duwal Markus R. Eisele Ana Finzel Pérez Andres Florez Michael Flossdorf Juan Fuxman Bass Avishai Gavish Miriam Gutschow Kevin Haigis Tim Heinemann Frank Stefan Heldt Heinrich Huber Junguk Hur 62 Title A method for parsing clinical outcomes and combining them with phopshoproteomic and genomic data for predicting drug efficacy: application in hepatocellular carcinoma Model-guided target identification for synergistic combination therapies in the DNA damage response pathway Modelling the dynamics at the interface between autophagy and apoptosis Controllability of protein interaction network identifies human disease genes Epstein-Barr virus Integrome and Infectome of the Human Genome by D-ViSioN (Detection of Integration of Virus(s) by SingletoN(s)) The human blood metabolome-transcriptome interface A dynamic view on cell-to-cell variability in protein levels is required to investigate the molecular mechanisms behind non-genetic resistance in TRAIL-induced apoptosis Characterization of genetic interaction classes in a metazoan reveals the modular organisation of genetic interactomes and suggests prioritizing strategies for combinatorial GWAS in human Concise Functional Enrichment of Ranked Gene Lists An Exploratory Analysis of Shared Genetic Effects between Brain Oscillations and DSM-IV Alcohol Dependence Systems Pharmacology: Enabling early quantitative decisions in pharma from start of lead identification to clinical trials Sequence Analysis of the mt-COXII Gene in Crassostrea virginica: Application to Human Disease Systemic analysis of expression data of glioblastoma multiforme patients OpenBEL - A platform for capture, integration, and application of biological knowledge Dosage of Dyrk1a shifts cells within a p21-cyclin D1 signaling map to control the decision to enter the cell cycle Systems analysis of cytokine-mediated multicellular cross-talk in normal and inflammatory conditions Microfluidic high-throughput platforms for live-cell imaging and transcriptional quantification in immune cells at single cell resolution Discovery of kinase motifs and prediction of target substrates using the ProPeL method Emerging landscape of oncogenic signatures across human cancers Decoding Network-Attacking Mutations in Cancer Sequencing and Analysis of the mt-ATP6 Gene in Crassostrea virginica In Situ Sequencing: A Multi-Plex Multi-Omic Multi-Scale Technology Approaches for Integrative Analysis of Pharmacology and Genome-wide Molecular Profiling Data Applied to a Large Scale Screen of Cancer Cell Lines Phylogenetic profiling reveals novel functional modules in the human genome Integration of multi-leveled -omics data for cancer network analysis. PK-PD Modelling of the Reverse Transcriptase Inhibitor Tenofovir and Quantification of its Prophylactic Efficacy against HIV-1 Infection Cell-to-Cell Communication Modulates the TNF-a and IL-6 Response to TLR4 Stimulation in Macrophages: A Computational Approach The role of p53 in metabolic regulation and response to cellular perturbations The mechanistic role of MYCN in driving cell cycle progression in Neuroblastoma T cell immune responses generate diversity through linear cell-fate progression rather than asymmetric cell divisions Human transcription factor network evolution and rewiring in disease Robust pattern formation in the Drosophila eye disc Multiple stimuli converge to influence transcription factor dynamics. Cytokine signaling in Alzheimer’s disease Robust DNA Repair through Collective Rate Control A multiscale model of influenza A virus infection that elucidates the treatment with direct-acting antivirals Modeling NOS3 signaling as markers for NO-bioavailability, ROS generation and cGMP-metabolism in cardiac pathophysiology Systems pharmacology analysis of drug-induced peripheral neuropathy No. 16 Day Wed 18 Wed 17 100 Tue Wed 19 Tue 20 21 Wed Tue 22 Wed 23 25 Tue Tue 24 Wed 26 Wed 27 28 4 Tue Wed Wed 6 Wed 5 Tue 29 Tue 30 31 32 7 2 Wed Tue Wed Tue Wed 33 34 35 Tue Wed Tue 36 Wed 37 38 39 Tue Wed Tue 40 41 42 43 44 45 Wed Tue Wed Tue Wed Tue 46 Wed 47 Tue International Conference on Systems Biology of Human Disease Presenter Martine A Jaworski Enes Karaboga S. Jordan Kerns Antje Kettelhake Christina Kiel Daniel Kirouac Bettina Knapp Ariel Kniss Fabian Konrath Anil Korkut Yan Kou Erika Kuchen Natasha Kumar Young Kwon Kasper Lage Anastasiya Lapytsko Dooyoung Lee Victor Li Jia-Ren Lin Carlos Lopez Roy Malka Paula A. Marin Zapata Steve Martin Linas Mazutis Ioannis N. Melas Dimitris E Messinis Aaron Meyer Miles Miller Kathryn Miller-Jensen Shekhar Mishra Janina Mothes Steven C. Neier Ian Overton John Paasch Christian Priesnitz Stephen Ramsey Andreas Raue Erzsébet Ravasz Regan Boston, MA Title A systems biology approach to the GABAergic contribution to Juvenile Myoclonic Epilepsy Application of S-pyocins to eradicate Pseudomonas aeruginosa biofilms Detecting and Remembering Gut Dysfunction in Biomimetic Microfluidic Devices with Living Bacterial Diagnostics Systems Biology Evaluation of Hepatocellular Carcinoma Metabolism Structure-energy-based predictions and network modelling of RASopathy and cancer missense mutations In silico identification of biomarker-optimized treatment strategies in HER2+ cancer Inter-individual effects mask the molecular signature of psoriasis and eczema Developing microfluidic and experimental tools to interrogate intracellular T cell signaling using a frequency response analysis approach Identification of new IkappaBalpha complexes by an iterative experimental and mathematical modeling approach Network models of signaling and drug response in melanoma Enrichment vectors identify glioblastoma multiforme patient subgroups and suggest molecular mechanisms and potential treatments MYCN and cellular decisions in neuroblastoma TASK-2 channels contribute to pH sensitivity of retrotrapezoid nucleus chemoreceptor neurons The Hippo Signaling Pathway Interactome Functional interpretation of genome-wide association signals in arrhythmia using protein networks of cardiac ion channels Auto-inhibition stabilizes delayed negative feedback in biochemical systems A systems biology model for the coagulation network in non-bleeding state describes baseline activity of clotting Molecular ties between the cell cycle and differentiation in embryonic stem cells A Quantitative Drug-Target Relationship Network for Probing Cellular Response and Poly-pharmacology of Small Molecular Kinase Inhibitors Exploring how cells commit to apoptotic or necrotic cell-death. In Vivo Volume and Hemoglobin Dynamics of Red Blood Cells Mathematical modelling of the regulation and degradation of the transcription factor EB A systems biology model of the coagulation network in bleeding state reveals differences in behaviors of biomarkers in response to perturbations. Barcoding thousands of single cells in a single tube by droplet microfluidics Identification of deregulated signaling pathways in Multiple Sclerosis based on gene expression data Construction of a drug-induced phosphoprotein/cytokine dataset in clinical samples for Multiple Sclerosis The AXL Receptor is a Sensor of Ligand Spatial Heterogeneity Single-cell pharmacokinetic modeling of a clinically approved nanoparticle allows prediction of nano-therapeutic drug delivery to tumor microvasculature and associated macrophages Analysis of single-cell secretion reveals a role for paracrine signaling in coordinating macrophage response to TLR4 stimulation Reversible Bile Transport : A Cholesterol trader in Human Body Dynamic variability of NF-kappaB signal transduction: A mechanistic model Novel multiplex analysis of protein complexes in signaling networks reveals signatures that distinguish T-cell responses to antigen A map of functionally coherent binding for Snail and Twist transcription factors in fly mesoderm development informs regulation of epithelial remodelling and oncogenic Notch A Targeted Therapy - So, what is it targeting? A Case Study of the On-Target Effect on Waldenstrom’s and the Off-Target Effect on Platelet Count and What One May Tell Us About the Other Metabolomic characterization and analysis of liver-specific functions of an in vitro fibrosis model Network topological characteristics aid in identifying causal genes for atherosclerosis The Effect of Higher-Order Receptor Clusters on TRAIL Induced Apoptotic Signaling Organ-specific stochastic phenotype switching is required for endothelial health June 17-19, 2014 No. 49 Day Tue 50 112 Wed Wed 51 52 Tue Wed 53 54 3 Tue Wed Tue 55 Tue 56 8 Wed Wed 57 58 Tue Wed 59 60 Tue Wed 61 62 Tue Wed 63 9 Tue Tue 10 64 73 Wed Wed Tue 70 Wed 65 66 Tue Wed 11 Tue 67 12 Tue Wed 1 Tue 68 69 71 Wed Tue Tue 48 Wed 72 Wed 74 Wed 75 76 77 Tue Wed Tue 63 Poster Abstracts Presenter Faisal Reza Edward Rietman Jens Roat Kultima Jenny Rouka Assieh Saadatpour Anne Sadewasser Theodore Sakellaropoulos MN Salleh Somponnat Sampattavanich Shimon Sapir Sibaji Sarkar Ulf Schmitz Amitabh Sharma Ashwini Kumar Sharma Sriganesh Srihari Steven N. Steinway Edward Stites Jette Strasen Madhuresh Sumit Jaeyoung Sung Jaeyun Sung Christian Tokarski Stepan Tymoshenko Lakshmi Venkatraman Anita Voigt Astrid Wachter Allon Wagner Chun-Chao Wang Ruisheng Wang Kathleen Wilkie Jennifer Wilson Victor Wong Levi Wood Xianfang Xia Nilgun Yilmaz Carissa L. Young 64 Title Treating monogenic human diseases with evolving computationally designed mutagenic triplex-forming and recombinagenic donor DNA molecules Topological Measures of Protein-Protein Interactions And Implications For Cancer Therapy MOCAT: a metagenomics assembly and gene prediction toolkit Identification of novel CD2AP SH3 interaction partners by combining crystallography, bioinformatic tools and peptide arrays Characterizing heterogeneity in leukemic cells using single cell gene expression analysis Characterization of the protein signatures of permissive vs. non-permissive influenza A virus infections in human host cells by quantitative proteomic analysis Training of signaling pathways to phosphoproteomic data via hybrid node/edge optimization using a mixed integer programming formulation Study on the Clinical Significance of c-FLIP and JUN-B Expression in Psoriatic Lesion by using Quantitative RT-PCR and Tissue Microarray Pulsatile FoxO3a translocation in response to growth factors is controlled jointly by AKT and ERK pathways Speech abnormalities in Parkinson’s disease reflect a major disturbance in sensorimotor servomechanisms Systems Biology Approach to Determine the Efficacy of Methylation by DNA-Methyl Transferase1 in Cancer Cells Comprehensive identification of cooperative microRNA target regulation through an integrative workflow Diseaseome 2.0: Uncovering disease-disease relationships through the human interactome Proximity of metabolic and cancer causing genes in the genome leads to metabolic remodeling in cancers. Harnessing DNA Damage Repair Pathways in Breast Cancer Therapy: a Synthetic Lethality Paradigm Network modeling reveals key features of epithelial-to-mesenchymal transition dynamics Ras mutant-specific response to upstream inhibition TGFb Signaling in Living Cells Delineating G-protein coupled receptor (GPCR)-linked oscillatory signaling mechanisms through single cell microfluidic analysis and computational modeling A new stochastic kinetics approach to quantitative description of intracellular networks coupled to hidden cell environment Global metabolic interaction network of the human gut microbiota Hepatic Response to Refeeding - From Ketone Bodies to Lipids Reconstruction and in silico analysis of metabolism in the apicomplexan parasite Toxoplasma gondii Systems-biology of Poly(I:C) induced apoptosis signaling Assessment of the human gut microbiome in response to temporal and colorectal cancer-associated variability High-Throughput Proteomic and Transcriptomic Data Integration based on MS/MS and RNA-Seq Data using Prior Pathway Knowledge Drug efficacy and side effects are strongly associated with its ability to reverse gene expression abnormalities in a mouse model of dyslipidemia A dynamic regulatory circuit in single breast epithelial cells and basal-like premalignancies Network-based association of hypoxia-responsive genes with cardiovascular diseases Classifying Cancer Types for Treatability Using PPI Network Structure Network algorithms dicover dysregulated pathways in RNAi screens Cell-to-cell variability in NF-κB activation and chromatin at the HIV promoter contribute to noise in the reactivation of latent HIV In Vivo Systems Analysis Identifies Cytokine Drivers of Neurodegeneration in Alzheimer’s Disease Exploring the cellular responses to short pulses of TNF A Multiscale-Modeling Approach Towards Understanding the Biological Response to Stress Elucidating Mechanisms of Liver Metastasis: an All-Human Microphysiological Model to Investigate Disease Progression & Therapies No. 78 Day Wed 79 Tue 99 80 Tue Wed 81 Tue 82 Wed 83 Tue 84 Wed 85 Tue 86 Wed 87 Tue 88 Wed 89 Tue 90 Wed 91 Tue 92 93 13 94 Wed Tue Tue Wed 95 Tue 96 97 98 Wed Tue Wed 14 101 Wed Tue 102 Wed 103 Tue 15 Tue 104 105 106 107 Wed Tue Wed Tue 108 Wed 109 110 Tue Wed 111 Tue International Conference on Systems Biology of Human Disease #1 : Analysis of single-cell secretion reveals a role for paracrine signaling in coordinating macrophage response to TLR4 stimulation Qiong Xue1 , Yao Lu1 , Markus Eisele1,2 , Nafeesa Khan1 , Rong Fan1 , Kathryn Miller-Jensen1 1 2 Yale University, New Haven, CT University of Stuttgart, Stuttgart, Germany Cell populations produce reliable biological responses despite significant levels of cellto-cell heterogeneity. These biological responses often involve intermediate extracellular signals in which cells secrete and respond to the same factor. The propagation of intermediate signals by extracellular signaling has a significant impact on the collective cellpopulation response, but the contribution of autocrine versus paracrine signaling remains difficult to analyze. Here we combined multiplexed, nanowell single-cell secretion measurements with cell-population data to distinguish the role of paracrine signaling in shaping the inflammatory cytokine secretion profile in a monocytic cell line (U937) and primary human monocyte-derived macrophages (MDMs) in response to toll-like receptor 4 (TLR4) stimulation with lipopolysaccharide (LPS). Loss of paracrine signaling upon single-cell isolation in nanowells significantly reduced secretion of a subset of LPS-stimulated cytokines, including interleukin-6 (IL-6) and IL-10, in both U937 and primary human MDM cells. Using graphical Gaussian modeling of LPS-stimulated single-cell data, we identified tumor necrosis factor-α (TNF-α) and IL-1β as central nodes in the LPS-stimulated paracrinesignaling network, which was further validated experimentally. In both U937 and MDM single cells, a small fraction of cells accounted for the majority of the total TNF-α output in the cell population, but paracrine signaling from this small subset amplified secretion and reduced variability in the downstream dependent signals. Our results demonstrate that a small subpopulation of high-responding cells appears to drive the innate immune response in macrophages through paracrine signaling. Thus, paracrine signaling results in a coordinated population response that is absent in isolated single cells, underscoring the importance of inter-cellular signaling in determining cellular behaviors. Boston, MA June 17-19, 2014 65 Poster Abstracts #2 : Approaches for Integrative Analysis of Pharmacology and Genome-wide Molecular Profiling Data Applied to a Large Scale Screen of Cancer Cell Lines Sara Dempster, Jonathan Dry AstraZeneca Pharmaceuticals LP, Waltham, MA AstraZeneca has partnered with the Wellcome Trust Sanger Institute to screen the AstraZeneca oncology portfolio across a diverse panel of 1000 cancer cell lines. The cell lines have been profiled by gene expression arrays, whole exome sequencing, and array CGH. The dataset provides a unique opportunity to identify novel molecular markers of drug sensitivity and resistance in pre-clinical models. We are integrating the datasets to identify novel molecular markers to inform hypotheses for patient segmentation. For example, one goal is to identify molecular features that are associated with differences in specificity and efficacy of AZ compounds that target the same pathways. Here, we focus on the application of methods that will support these goals. Because there are hundreds of thousands of molecular features, we need to reduce the dimensionality by grouping features likely to have redundant functional impact in order to improve the power of our analysis. To address this challenge, we are exploring approaches that leverage known biological pathways and networks such as HotNet and NBS as well as unsupervised clustering approaches such as iCluster Plus. We will present different approaches for applying the HotNet algorithm to the data, show initial results, and propose next steps. 66 International Conference on Systems Biology of Human Disease #3 : Developing microfluidic and experimental tools to interrogate intracellular T cell signaling using a frequency response analysis approach Ariel Kniss1 , Loice Chingozha2 , Hang Lu1,2 , Melissa Kemp1 1 The Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University 2 School of Chemical and Biomolecular Engineering, Georgia Institute of Technology Introduction: T cells are part of the adaptive immune response that recognize and fight pathogens in the body. T cell activation is a complex process resulting in the induction of a proliferative response, encoded by the dynamics of multiple signaling cascades and reliant on second messenger molecules such as calcium and hydrogen peroxide. Frequency response analysis (FRA), developed in control engineering, has been previously allowed for inference of cellular signaling network structure but has been experimentally difficult to apply to suspension T cells. We have designed a FRA experimental and computational platform for future use in determining the dominant feedback controls present in T cell signaling. Materials and Methods: To develop the experimental precision required for this approach, we use a microfluidic device capable of applying an oscillatory chemical cue as an input signal while enabling visualization of T cells through time. In this work we apply H2 O2 at a range of driving frequencies to Jurkat T cells while monitoring cytoplasmic calcium concentration. We then perform spectral analysis of the single cell calcium traces to determine the output signal response of the system. Results: Calcium signaling is found to respond differently to 100µM H2 O2 at specific driving frequencies of 16.7mHz (1 min period), 8.3mHz (2 min period), and 2.8mHz (6 min period). With the single cell spectral analysis, subpopulations of cells emerge in response to the same driving frequency and between different experiments. Cells are entrained to frequencies below 8.3mHz but attenuate H2 O2 signals above 16.7mHz. Conclusion: With the use of microfluidic technology and initial spectral analysis, we have determined that T cells have a cutoff frequency between 1 and 2 minutes when interrogated with oscillatory H2 O2 stimulus. This suggests downstream transcriptional effects are dependent on the frequency of environmental cues, with high frequency signals filtered out as potential noise. Boston, MA June 17-19, 2014 67 Poster Abstracts #4 : Dosage of Dyrk1a shifts cells within a p21-cyclin D1 signaling map to control the decision to enter the cell cycle Jia-Yun Chen1,2 , Jia-Ren Lin1,2 , Feng-Chiao Tsai1 , Tobias Meyer1 1 2 Department of Chemical and Systems Biology, Stanford University, Stanford, CA Laboratory of Systems Pharmacology, Harvard Medical School, Boston, MA Mammalian cells have a remarkable capacity to compensate for heterozygous gene loss or extra gene copies. One exception is Down syndrome (DS) where a third copy of chromosome 21 mediates neurogenesis defects and lowers the frequency of solid tumors. Here we combine live-cell imaging and single-cell analysis to show that increased dosage of chromosome 21-localized Dyrk1a steeply increases G1 cell cycle duration through direct phosphorylation and degradation of cyclin D1 (CycD1). DS-derived fibroblasts showed analogous cell cycle changes that were reversed by Dyrk1a inhibition. Furthermore, reducing Dyrk1a activity increased CycD1 expression to force a bifurcation, with one subpopulation of cells accelerating proliferation and the other arresting proliferation by co-stabilizing CycD1 and the CDK inhibitor p21. Thus, dosage of Dyrk1a repositions cells within a p21-CycD1 signaling map, directing each cell to either proliferate or to follow two distinct cell cycle exit pathways characterized by high or low CycD1 and p21 levels. 68 International Conference on Systems Biology of Human Disease #5 : Microfluidic high-throughput platforms for live-cell imaging and transcriptional quantification in immune cells at single cell resolution Loice Chingozha1 , Cheng Zhu2,3,4 , Hang Lu1,3 1 Georgia Institute of Technology, School of Chemical and Biomolecular Engineering, Atlanta, GA Georgia Institute of Technology, School of Mechanical Engineering, Atlanta, GA 3 Interdisciplinary Program in Bioengineering, Georgia Institute of Technology, Atlanta, GA 4 Wallace H. Coulter Department of Biomedical Engineering, Georgia Tech and Emory University, Atlanta, GA 2 Analyzing immune T cell response to stimulation requires assays that are highly sensitive, provide real-time measurements, and enable single-cell analysis. The inherent cell heterogeneity that can have pathological consequences necessitates assays that allow for single-cell resolution in a high-throughput manner. Here we developed a microfluidic platform for (1) single cell handling, (2) performing cell dynamic stimulation with simultaneous live-cell imaging, (3) immunofluorescence for surface marker expression and cytokine secretion characterization, and (4) single-cell Fluorescence In Situ Hybridization (smFISH) for quantification of single mRNA transcripts at a single cell resolution for hundreds of cells in a single device with reduced reagent use. smFISH is a quantitative method to characterize mRNA transcripts in cells; it uses multiple short, complementary oligonucleotides, and therefore is accurate to single transcript level. Our device enables cell stimulation via soluble stimulation with temporal patterns or stimulation with surfaceanchored molecules, while simultaneously performing fluorescence-based live-cell imaging. Subsequent downstream characterization of RNA transcription using smFISH and immunofluorescence characterization at single cell resolution enables us to correlate early time events with the subsequent late time functional response in the same cell with a throughput of a thousand cells per chip. Therefore, this technique will contribute greatly to systems understanding cellular functional response under system perturbations such as immunological cell response to viral transcription in infected hosts. Overall, our design provides a tool for studying stochasticity underlying gene expression, and will enable advancement in single-cell immunophenotyping and transcriptomics. Boston, MA June 17-19, 2014 69 Poster Abstracts #6 : Systems analysis of cytokine-mediated multicellular cross-talk in normal and inflammatory conditions Kelly W. Chen1 , Jason Velazquez1 , Linda G. Griffith1 , Rebecca Carrier2 , Douglas A. Lauffenburger1 1 2 Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA Department of Chemical Engineering, Northeastern University, MA The human gut is the largest immune organ in the body. Intestinal homeostasis is tightly regulated by the coordinated actions of a multitude of cell types, including enterocytes, goblet cells and immune cells. However, a quantitative understanding of how these cellular constituents communicate and integratively contribute to overall tissue function is lacking. Given the complexity of intercellular cross-talk, conventional approaches are no longer sufficient, and improved understanding necessitates multivariate experimental design and systems-level analysis. To this end, we have developed an immune-competent human intestinal model, which incorporates representative cellular components of the mucosal environment (enterocyte, goblet cells, macrophages/ dendritic cells). To investigate the contribution of each cell type to the overall cytokine milieu of the microenvironment, cellular composition in the intestinal model was varied systematically (e.g., +/-goblet cells, +/- immune cells) under normal and perturbed conditions (e.g., LPS, microbes, drugs). The secreted signals (cytokines, chemokines, growth factors) were measured using multiplex assays. Multivariate modeling techniques (e.g., partial least-squares regression) were used to quantitatively relate soluble signaling profiles to tissue-level phenotypes and functions (e.g., barrier function, mucus production, enterocyte differentiation), thereby enabling the identification of key soluble mediators that contribute to divergent cell responses (normal versus inflamed). The predictive models generated from this study can guide hypothesis generation regarding cytokine-mediated multicellular interactions and their implications for tissue-level phenotypes/functions. Our experimental approach, which combines physiological relevant in vitro culture platforms and computational strategies, has broad applicability in advancing fundamental research in mucosal immunology and drug development. 70 International Conference on Systems Biology of Human Disease #7 : In Situ Sequencing: A Multi-Plex Multi-Omic MultiScale Technology Evan Daugharthy1,2,3 , Je-Hyuk Lee1,2 , Richard Terry2 , Jonathan Scheiman1,2 , Paul Tillberg4,5 , Fei Chen4,5 , Benjamin Pruitt2 , Brian Turczyk2 , Reza Kalhor1,2 , John Aach1 , Ed Boyden4,5 , George Church1,2,4 1 2 3 4 5 Department of Genetics, Harvard Medical School, Boston MA Wyss Institute for Biologically Inspired Engineering, Harvard University, Boston MA Department of Systems Biology, Harvard Medical School, Boston MA MIT Media Lab, Cambridge MA Departments of Biological Engineering Brain and Cognitive Sciences, Cambridge MA Biological systems exhibit meaningful spatial heterogeneity over 15 orders of magnitude in scale and in an even larger space of genetic information and material compositions. Unfortunately, in situ methods such as fluorescent in situ hybridization and immunohistochemistry are low-throughput, have low-multiplexity, and have quantitativity and spatial measurement scales limited by the physics of light and electronic photosensitivity. We have recently demonstrated massively multiplex fluorescent in situ RNA sequencing, in which stably cross-linked cDNA amplicons are sequenced within a biological sample. Using 30-base reads from 8102 genes in situ, we examined RNA expression and localization in human primary fibroblasts with a simulated wound-healing assay. Moreover, our platform enables massively parallel detection of genetic elements; simultaneous measurements of the genome, transcriptome, and proteome; and programmable spatial analysis scales from single cell to whole organism. As a tool for systems biologists, these quantitative measurements can illuminate the molecular-genetic foundation of human infirmities to be addressed with genome engineering and other precisely targetted therapies. Boston, MA June 17-19, 2014 71 Poster Abstracts #8 : Enrichment vectors identify glioblastoma multiforme patient subgroups and suggest molecular mechanisms and potential treatments Yan Kou1 , Qiaonan Duan1 , Neil Clark1 , Jann Sarkaria2 , James Gallo1 , Avi Ma’ayan1 1 Department of Pharmacology and Systems Therapeutics, Systems Biology Center New York, Icahn School of Medicine at Mount 2 Department of Radiation Oncology, Mayo Clinic College of Medicine, Rochester, MN 55905 Glioblastoma multiforme (GBM) is the most common and aggressive malignant primary brain tumor. Although treated with a combination of chemotherapy, radiation, surgery and anti-tumor drugs, the typical survival for GBM patients is less than 15 months. Furthermore the diverse response of individuals to the various treatment regimens poses intriguing questions for biologists about the underlying molecular mechanisms. Four molecular subtypes of glioblastoma have been identified by clustering patient tumor’s signature genes expression. Nonetheless, these subtypes neither suggest underlying molecular mechanisms nor provide direct suggestion to the most effective potential treatments that would benefit patients belonging to a certain subtype. We developed a novel patient classification method by systematically integrating data from large ongoing cancer genomic projects such as The Cancer Genome Atlas, Cancer Genome Project as well as data that we processed from ENCODE, ChEA, the Roadmap Epigenomics, and Mouse Genome Informatics Mammalian Phenotypes. By applying gene set enrichment analyses on the up-regulated genes obtained from each glioblastoma patient, we constructed an enrichment vector clustering of patients. We show that classification of GBM patients based on the enrichment vectors is able to identify 9 clusters with 3 having significant difference in clinical outcome. Next, we investigated the correlation between the patient subgroups and 40 glioblastoma cell lines and we found cell lines that share similar enrichment vector profiles with specific patient clusters. Finally, using the drug sensitivity data from CGP we created a tripartite network connecting patient clusters, glioblastoma cell lines and drugs to schematically reveal potential drugs that can benefit clusters of cancer patients. Without any prior knowledge, we predicted that bleomycin, a glycopeptide inhibitor currently undergoing Phase I trial, as a drug that would benefit the patient group with the poorest prognosis. The tripartite network also suggest the drugs NVP-TAE684 and OSI906 studied in other cancer types as potentially effective for specific glioblastoma patient clusters. 72 International Conference on Systems Biology of Human Disease #9 : A Quantitative Drug-Target Relationship Network for Probing Cellular Response and Poly-pharmacology of Small Molecular Kinase Inhibitors Jia-Ren Lin, Marc Hafner, Mario Niepel, Gabriel Berriz, Peter Sorger Laboratory of Systems Pharmacology, Harvard Medical School Systems pharmacology has been thrived over the past decade, given the huge amount of data gather from large-scale drug screens and functional genomics studies (Hopkins, 2008). Furthermore, the accumulated drug-target interactions has been summarized in databases like DrugBank and been broadly used in drug discovery. However, the lack of quantitative data sources or network tools hinders the overall understanding on drug action in network pharmacology. To fulfill this need, we constructed a computable drug-target networks using the aggregated information from PubChem bioassay database (Wang et al., 2013). This focused data-set contained around 200 drugs using in cancer therapy. From the network analysis of quantitative target-drug interactions, we have been able not only to identify different sub-clusters among drugs with the same nominal targets but also reveals the underlying biological networks of different drugs. In addition, we used these information to compute the specificity or “dirtiness” of drugs and confirmed these predictions experimentally using KinomeScan. Finally, the similarity of drugs from their target networks are correlated with the drug sensitivity matrix in NCI60 panel, suggesting the network information can be used to predict drug action in vivo. Altogether, our computational platform here provides researchers with a useful tool to probe poly-pharmacology in cancers and a comprehensive view of mutli-drugs multi-target networks in systems biology. 1. Hopkins, A. L. (2008). Network pharmacology: the next paradigm in drug discovery. Nature chemical biology, 4(11), 682–90. doi:10.1038/nchembio.118 2. Wang, Y., Suzek, T., Zhang, J., Wang, J., He, S., Cheng, T., . . . Bryant, S. H. (2013). PubChem BioAssay: 2014 update. Nucleic acids research, (September 2013), 1–8. doi:10.1093/nar/gkt978 Boston, MA June 17-19, 2014 73 Poster Abstracts #10 : Exploring how cells commit to apoptotic or necrotic cell-death. Michael Irvin, Patrice Wagner, Sandra Zinkel, Carlos Lopez Vanderbilt University School of Medicine Necrosis has recently emerged as a programmed cell-death alternative to cellular apoptosis. Programmed necrosis is important in the pathology of a number of human diseases including myocardial infarct, inflammatory bowel diseases, stroke, and neurodegeneration. Death receptor mediated signaling can induce either apoptotic or necrotic cell death and thus represents an ideal system to understand the mechanistic origins of cellular decisions processes. This work uses novel mathematical modeling approaches and experimental data, to explore mechanistic hypotheses about cell death decisions between apoptosis or necrosis. We explore on results from the Zinkel lab that demonstrate that the pro-apoptotic protein Bid, inhibits and modulates Rip1-driven programmed necrosis. Given that Bid and RIP1 function within a complex network of biological signals, we explore the systems-level behavior of protein interactions that could regulate necrosis or apoptosis signals. We generate mathematical models, calibrated to experimental data, which describe biochemical interactions that can modulate apoptosis or necrosis outcome. Our apoptosis-necrosis reaction model (ANRM) extends previous apoptosis mathematical models with new necrosis pathway interactions to describe a comprehensive cell death mechanistic framework. ANRM can simulate cellular processes that lead to either apoptosis or necrosis cell fates based only on initial protein concentrations. The ANRM biochemical interaction topology has been calibrated to experimental data using parameters from multiple cells including Jurkat, and Hematopoietic Progenitor cells. We find that the best fits to experimental data occur when a regulatory step for Bid is accentuated. In our work, anti-necrotic activity of Bid required regulation with kinetic rate four orders of magnitude different than generic values thus implying, from a modeling perspective, that this interaction is fundamental to cell commitment to either form of cell death. 74 International Conference on Systems Biology of Human Disease #11 : Construction of a drug-induced phosphoprotein/cytokine dataset in clinical samples for Multiple Sclerosis Vaia Pliaka1,2 , Dimitris E Messinis1,2,3 , Theodore Sakellaropoulos3 , Ekaterina Kotelnikova4 , Tomas Olsson5 , Jesper Tegner6 , Roland Martin7 , Dimitris Tzeranis1,3 , Friedemann Paul8 , Elena Schwartz9 , Ilya Mazo9 , Sophia Stamatatou1 , Mar Masso10 , Albert Zamora10 , Pablo Villoslada4 , Leonidas G Alexopoulos1,3 , Julio Saez-Rodiguez11 , Marti Bernardo-Faura11 , Ioannis N Melas1,3,11 , Jose Manuel Mas12 , Laura Artigas12 1 ProtATonce Ltd, Athens, Greece equal contributor 3 National Technical University of Athens, Greece 4 IDIBAPS – Hospital Clinic of Barcelona, Spain 5 Department of Neurology, Karolinska Institutet, Sweden 6 Unit of Computational Medicine, Karolinska Institutet, Sweden 7 University of Zurich, Switzerland 8 NeuroCure Clinical Research Center, Charité University Medicine Berlin, Berlin, Germany 9 Ariadne Diagnostic, Rockville, MD, US 10 Bionure Farma SL, Barcelona, Spain 11 European Molecular Biology Laboratory, European Bioinformatics Institute, UK 12 Anaxomics Biotech, Barcelona, Spain 2 Multiple Sclerosis (MS) is an autoimmune disease that affects the brain and spinal cord. An estimated 2,500,000 people around the world have MS and there is not yet a cure for the disease. Even though significant progress is currently being made in MS research, the pathogenesis of the disease has not been comprehensively understood. A great number of pathological mechanisms responsible for the disease have been described, involving genes & proteins altering many processes & pathways. By understanding how current MS therapies work in biological networks, more effective therapies can be designed. On this front, the CombiMS consortium (combims.eu) is developing computational and experimental tools to improve the therapeutic options of MS in the future. A milestone in this consortium is to evaluate how current and “promising” MS drugs work at the signaling level in different patient populations. Peripheral blood mononuclear cells (PBMCs) from 255 donors were collected by several European medical centers. The PBMCs were plated in 96 well plates and 20 stimuli & drugs were applied. Using custom multiplex assays, 17 phosphoproteins were measured at 5 & 25 minutes and 22 cytokines at 24h post stimulus. To ensure the highest possible data quality, a kit was developed to include all reagents needed in order to isolate PBMCs from 1 donor and then plate, stimulate and lyse them. Sample collection controls were applied in the stimuli set to evaluate Boston, MA June 17-19, 2014 75 Poster Abstracts errors in sample processing and 2 custom xMAP bead sets were used to evaluate errors and instrument measuring variability in the bead-based ELISA procedure. Phosphoproteomic and cytokine data will be combined with SNP data and clinical profiles (i.e. responders, non-responders, therapeutic intervention) in a computational framework which will help to understand MS more thoroughly and systematically. As a first step, the phosphoproteomic dataset will be used for the construction of a detailed map of the signaling pathway differences between MS and healthy donors, which can help generate a model of MS pathogenesis and improve our understanding of the disease. 76 International Conference on Systems Biology of Human Disease #12 : Single-cell pharmacokinetic modeling of a clinically approved nanoparticle allows prediction of nano-therapeutic drug delivery to tumor microvasculature and associated macrophages Miles Miller1 , Suresh Gadde2 , Christina Pfirschke1 , Rainer Kohler1 , Ashley Laughney1 , Mikael Pittet1 , Omid Farokhzad2 , Ralph Weissleder1 1 MGH Center for Systems Biology, Harvard Medical School Laboratory of Nanomedicine and Biomaterials, Brigham and Women’s Hospital (BWH), Harvard Medical School 2 Nanoparticle encapsulation for targeted and controlled therapeutic delivery holds promise for a variety of drugs, yet it is has been generally difficult to ascertain at high resolution how such therapeutic nanoparticles (TNP) function in vivo, especially in patients. Magnetic nanoparticles (MNP) such as the clinically approved ferumoxytol (Feraheme) represent attractive magnetic resonance imaging agents to select and monitor patients for TNP-based drug treatment, yet the compartmental tumoral distribution and possible co-localization with TNP is unknown. Here we used single cell resolution intravital imaging to address both questions. We visualize MNP in tumor microvasculature and show sustained uptake into tumor-associated-macrophages within minutes. When compared against a model TNP (poly(D,L-lactic-co-glycolic acid)-b-poly(ethylene glycol); PLGA-bPEG) that is based on several clinically tested therapeutic formulations, we show excellent correspondence between the two nanoparticles in the microvasculature and within tumor associated macrophages. Computational non-linear finite-element modeling of NP transport enables predictive quantification of TNP distribution based on imaging data, and is used to identify key parameters governing overall intratumoral NP accumulation. These approaches yield insight into in vivo drug action, especially as it relates to enhanced permeability and retention (EPR) effects thought to be critical for TNP efficacy. Furthermore, the studies form the basis for clinical approaches to select patients into TNP trials based on MNP imaging. Boston, MA June 17-19, 2014 77 Poster Abstracts #13 : TGFb Signaling in Living Cells Jette Strasen1 , Uddipan Sarma2 , Stefan Legewie2 , Alexander Loewer1 1 2 Max-Delbrueck-Center (MDC), Berlin Institute for Medical Systems Biology (BIMSB), Berlin, Germany Institute of Molecular Biology (IMB), Mainz, Germany The TGFβ pathway is a multi-functional signaling system regulating cellular processes ranging from proliferation and migration to differentiation and cell death. Upon ligand binding, SMAD proteins translocate to the nucleus and activate numerous genes. While many components of the TGFβ pathway have been identified, we are still challenged to understand how its activation is translated into distinct cellular responses. We hypothesize that the information is encoded in stimulus- and context-specific dynamics of SMAD translocation. We therefore aimed to quantify long-term SMAD dynamics and to identify the network interactions that shape them. As the cellular response to a given stimulus often varies even in genetically identical cells, we focused on measuring pathway activity in single cells. By combining fluorescent reporter cell lines with live-cell microscopy and automated image analysis, we monitored the cytoplasmic to nuclear translocation of SMADs with high temporal and spatial resolution in hundreds of individual cells. Our experiments demonstrated that pathway activity can be divided into a first synchronous phase of SMAD translocation, followed by a second response that shows high variability in the extent and duration of nuclear accumulation. The amplitude of the first response is sensitive to low TGFβ concentration, while the duration of the second response encodes information at high ligand doses. Perturbation experiments indicated that both receptor internalization or degradation and transcriptional feedbacks contribute to shaping the dynamic response. We are now interested in further characterizing memory in the system to understand how information is encoded. To guide this analysis, we are developing a mathematical model that reflects signaling dynamics both at the population and single cell level. Our study will provide a deeper understanding of the molecular mechanisms regulating TGFβ signaling and open opportunities to modulate it in diseased cells. 78 International Conference on Systems Biology of Human Disease #14 : Systems-biology of Poly(I:C) induced apoptosis signaling Lakshmi Venkatraman, Joel Beaudouin, Roland Eils Department for Bioinformatics and Functional Genomics,German Cancer Research Centre, BioQuant, Heidelberg, Germany Cellular apoptosis is one of the immune response mechanisms used by cells to counteract the presence of foreign DNA/RNAbrought by viruses1 . Although inflammatory response to virus entry is well documented, a comprehensive study of the apoptosis signaling initiated by virus is currently absent. In this study using a combination of biochemical methods and live cell imaging we aim to construct a mathematical model of virus-induced apoptosis signaling. Addition of the dsRNA mimetic, polyinosinic:polycytidylic acid (poly(I:C)), both dose and time dependently increased apoptosis in HeLa cells. However we observed cellular variability in apoptotic response, with only a fraction of cells dying and with a strong variability in the timing. Overexpression of Toll-like receptor3 (TLR3) or pretreatment with the protein synthesis inhibitor, cycloheximide increased sensitivity to poly(I:C)-induced apoptosis. To analyze the cellular variability in apoptosis, fluorescent caspase probes for the initiator caspase-8 and the effector caspase-3 are imaged in living cells over time. Interestingly in cells treated with poly(I:C) a time-lag between caspase-8 activation and caspase-3 activation was observed. Presence of thresholds in the signaling pathway leading to caspase-8 activation could potentially explain cellular variability in apoptotic response. To analyze the cause of this threshold, upstream signaling molecules like TLR3 and mitochondrial membrane protein MAVS are perturbed in the presence of poly(I:C). Kinetics obtained from experiments are further utilized to create a mathematical model of casp8 activation and apoptosis. A quantitative model detailing the apoptosis signaling initiated after poly(I:C) induction would be beneficial in analyzing the potential role of poly(I:C) in targeted cancer therapy2 . 1. Trends in Microbiology (1999), 7,160. 2. Nature Biotechnology (2014), 32,182. Boston, MA June 17-19, 2014 79 Poster Abstracts #15 : A dynamic regulatory circuit in single breast epithelial cells and basal-like premalignancies Chun-Chao Wang1 , Sameer Bajikar1 , Leen Jamal1 , Kristen Atkins2 , Kevin Janes1 1 2 Department of Biomedical Engineering, University of Virginia, Charlottesville, Virginia Department of Pathology, University of Virginia, Charlottesville, Virginia Basal-like carcinoma is a subtype of breast cancer that is characterized by poor prognosis and high intratumor heterogeneity. In basal-like breast epithelia, we have identified two anti-correlated gene-expression programs that arise among single extracellular matrix (ECM)-attached cells during organotypic 3D culture. The first program contains multiple TGFβ-related genes including TGFBR3, and its heterogeneous induction is critical to suppress ductal branching. The second program contains JUND together with the basal-like marker, KRT5. Homogenizing JUND expression in single cells leads to 3D acini with bridged lumina that are similar to cribiform ductal carcinoma in situ. TGFBR3 and JUND form a circuit that is interconnected via four negative-feedback loops, which give rise to spontaneous damped oscillations in 3D culture. The TGFBR3–JUND circuit is remarkably conserved in basal-like premalignant lesions that heterogeneously express KRT5, suggesting that asynchronous circuit dynamics are active in this patient subset. We further show that the circuit is strongly dependent on ECM engagement, as detachment leads to a rewiring that is triggered by RPS6 dephosphorylation and maintained by juxtacrine signaling from tenascin C. Importantly, disruption of tenascin C substantially inhibits intraductal colonization of basal-like breast cancer cells in vivo. Breast tumor heterogeneity need not stem from partial basal-like differentiation and could instead reflect dynamic toggling of individual cells between expression states that are not cell autonomous. 80 International Conference on Systems Biology of Human Disease #16 : A method for parsing clinical outcomes and combining them with phopshoproteomic and genomic data for predicting drug efficacy: application in hepatocellular carcinoma Orfeas S. Aidonopoulos1,2 , Ioannis N. Melas3 , Theodore Sakellaropoulos1 , Leonidas G. Alexopoulos1 1 2 3 Dept. of Mechanical Engineering, National Technical University of Athens, Greece Dept. of Informatics and Telecommunications, National and Kapodistrian University of Athens, Greece European Molecular Biology Lab, European Bioinformatics Institute (EMBL-EBI), Cambridge, UK Systems biology has become an essential component of drug discovery, attempting to combine experimental data with computational modeling to capture different levels of cellular function (such as signaling, transcription, regulation) and integrate them in predictive models. These models are then used to best understand the drug mode of action (MOA), identify new targets and predict clinical drug efficacy and toxicity. In this study, we tried to identify signaling pathways related to drug efficacy in one of the most common malignancies worldwide, hepatocellular carcinoma (HCC). Particularly, gene expression data were collected for various HCC cell lines treated with anticancer compounds and a feature selection method was applied to identify genes predictive of the drugs’ clinical efficacy. For labeling each sample we constructed a graphical user interface that parses clinical trials databases for clinical outcomes containing the respective compound. Then, using the extracted data as an input to a pathway construction algorithm, we were able to infer signaling networks on the proteomic level that best fit the measured gene expression signatures. Finally, a high-throughput phosphoproteomic dataset was used to validate model predictions. Boston, MA June 17-19, 2014 81 Poster Abstracts #17 : Modelling the dynamics at the interface between autophagy and apoptosis Matthew Anderson, Virginie Betin, Robert Szalai, Jon Lane University of Bristol, Bristol, UK The dynamics involved in the cross-talk between apoptosis and autophagy has only recently come under the scrutiny of the scientific community, despite being key in the trade-off between cell death and survival, a crucial consideration for DNA-damaged cells. Using a comprehensive extant model of apoptosis as a framework, this study will combine known protein-protein interactions from the scientific literature with new experimental measurements of basal protein concentrations in order to construct a network model linking apoptosis and autophagy. Key elements of this relationship, such as mitochondrial fission and fusion, cytochrome c storage and release, and the numerous protein interactions involved in the cross-talk, will all be included in the completed model. Once tested experimentally and calibrated sufficiently, this model will be used to make predictions of system dynamics in order to answer key questions that will then be tested in the lab. The primary goal will be to assess how variation in mitochondrial structure influences the apoptotic and autophagic pathways, and how the activation thresholds of these pathways control the effects of targeted protein perturbations, the latter of which is crucial in the development of methods of disease treatment. 82 International Conference on Systems Biology of Human Disease #18 : Model-guided target identification for synergistic combination therapies in the DNA damage response pathway Andreas Raue, Shinji Oyama, Daryl Drummond, Birgit Schoeberl, Ozan Alkan Merrimack Pharmaceuticals Inc. Most commonly used anti-cancer therapies involve the systemic administration of antiproliferative and DNA damaging agents. Unfortunately, the dose of many chemotherapy agents leads to toxic side effects in many healthy tissues as well. In addition, systemic administration of such chemotherapy agents leads to accumulation of DNA mutations that may contribute to development of secondary cancers. The mechanism and pathways maintaining the genome stability and integrity are shared with the features of cell cycle regulation, which is deregulated in most human cancers. Loss of some elements of DNA damage response (DDR) or cell cycle pathways may be compensated by increased activity of other elements and therefore can cause resistance to DNA damaging therapies. Our goal is to gain mechanistic understanding of DNA damage response pathway topology and to develop a predictive computational model of the involved mechanisms. The model will help rationally design therapies targeting the DDR pathway. In particular, the discovery of synergistic and potentiating effects of drug combinations using standard-ofcare drugs, e.g. topoisomerase inhibitors such as doxorubicin and irinotecan/SN38, and existing or novel DDR targeting modulators/inhibitors are the main focus of this analysis. Furthermore, we hope to develop strategies that may overcome resistance. Finally, a model-based companion diagnostic strategy for the novel therapy is developed. Boston, MA June 17-19, 2014 83 Poster Abstracts #19 : Epstein-Barr virus Integrome and Infectome of the Human Genome by D-ViSioN (Detection of Integration of Virus(s) by SingletoN(s)) Gnanaprakash Balasubramanian1 , Barbara Hutter1 , Roland Eils2 , Benedikt Brors1 1 Computational Oncology Group, B080 Theoretical Bioinformatics, German Cancer Research Center (DKFZ), Heidelberg, Germany 2 B080 Theoretical Bioinformatics, German Cancer Research Center (DKFZ), Heidelberg, Germany Viruses are causal agents in various debilitating diseases including certain cancers. Globally 12% of the diagnosed malignancies have been associated with an oncovirus. Characterizing the viral infectious agent associated with cancer is of great importance in treating this dreadful disease. We have developed a method named D-ViSioN (Detection of Integration of Virus(s) by SingletoN(s)) that predicts viral integration sites based on Next Generation Sequencing. Our method also gives an overview of the associated viral infectome. D-ViSioN can be applied as a cancer pathogen discovery tool to decipher novel viral associations. We have applied this method to the whole genome sequencing data of Epstein Barr virus (EBV) transformed immortalized lymphoblastic cells. We have successfully obtained the EBV integrome and infectome of the cell line. Our method correctly detects the experimentally known integration sites of EBV. Genes affected by viral integration are involved in various basic cell processes like cell cycle control, cytoskeleton remodeling, vesicular traffic and cell communication. Thus, D-ViSioN gives a glimpse of the process of cell immortalization and might give valuable clues to decipher the phenomenon of oncogenic transformation. 84 International Conference on Systems Biology of Human Disease #20 : The human blood metabolome-transcriptome interface Joerg Bartel1 , Jan Krumsiek1 , Katharina Schramm2 , Jerzy Adamsky3 , Christian Gieger4 , Christian Herder5 , Annette Peters6 , Wolfgang Rathmann7 , Michael Roden5 , Konstantin Strauch4 , Karsten Suhre8,9 , Gabi Kastenmüller8 , Holger Prokisch2 , Fabian Theis1,10 1 Institute of Computational Biology, Helmholtz Zentrum München, 85764 Neuherberg, Germany Institute of Human Genetics, Helmholtz Zentrum München, 85764 Neuherberg, Germany 3 Institute of Experimental Genetics, Helmholtz Zentrum München, 85764 Neuherberg, Germany 4 Institute of Genetic Epidemiology, Helmholtz Zentrum München, 85764 Neuherberg, Germany 5 Institute for Clinical Diabetology, German Diabetes Center, 40225 Düsseldorf, Germany 6 Institute of Epidemiology II, Helmholtz Zentrum München, 85764 Neuherberg, Germany 7 Institute for Biometry and Epidemiology, German Diabetes Center, 40225 Düsseldorf, Germany 8 Institute of Bioinformatics and Systems Biology, Helmholtz Zentrum München, 85764 Neuherberg, Germany 9 Department of Physiology and Biophysics, Weill Cornell Medical College in Qatar, P.O. BOX 24144, Doha, Qatar 10 Department of Mathematics, TU Munich, D-80333 Munich, Germany 2 Simultaneous analysis of cross-sectional multi-omics data from large population studies enables a comprehensive characterization of subtle regulatory interactions on a physiological scale. In this study, we systematically analyzed the relationship between fasting serum metabolomics and whole blood transcriptomics data from 712 individuals of the German KORA F4 cohort. Correlation-based analysis identified significant associations between 522 transcripts and 114 metabolites summarized in an integrated network, which we subsequently call the ‘human blood metabolome-transcriptome interface’ (BMTI). Integration of these associations with the most recent genome-scale human metabolic network revealed a clear tendency towards higher correlations between metabolites and mRNAs which are closely connected in a pathway. Moreover, functional category-based aggregation of network nodes enables the interpretation of the large amount of inferred cross-omics associations, and allowed us to further confirm the biological validity of the BMTI. Inspection of subpathways of the resulting aggregated network uncovered both well-known mechanisms within a metabolic pathway, like the ‘carnitine-shuttle’, and novel information on the interplay between different biological processes like the immune system and monoacylglycerols. In a final step, we mapped associations to serum LDL, HDL and triglycerides onto the BMTI and the aggregated network, thereby identifying highly co-regulated network regions and differentially affected biological processes. Boston, MA June 17-19, 2014 85 Poster Abstracts We constructed a comprehensive gene-metabolite network derived from human whole blood transcriptomics. Results from association analysis with intermediate phenotypes like LDL, HDL and triglycerides demonstrated the utility of this network for future investigations in combination with disease-related phenotypes. 86 International Conference on Systems Biology of Human Disease #21 : A dynamic view on cell-to-cell variability in protein levels is required to investigate the molecular mechanisms behind non-genetic resistance in TRAIL-induced apoptosis François Bertaux, Szymon Stoma, Dirk Drasdo, Gregory Batt INRIA Paris-Rocquencourt Selective triggering of extrinsic apoptosis in cancer cells is a promising therapeutic strategy. But it faces many challenges, such as non-genetic resistance mechanisms, either natural (fractional killing) or acquired (decrease of efficiency as treatment is repeated). Recent single-cell, time-lapse, molecular-level measurements improved our quantitative understanding of extrinsic apoptosis: a better knowledge of its biochemistry was acquired and the crucial role of cell-to-cell differences in protein levels in determining cell fate variability was revealed. This knowledge was successfully formalized into mathematical models. But we still miss a comprehensive and quantitative picture of the molecular events allowing for cell survival and driving the evolution of resistance in surviving cells. Such understanding could lead to improved timing and dosage of drug administrations and reveal molecular targets to decrease resistance. Here we argue that mathematical models of extrinsic apoptosis used so far are fundamentally limited to investigate the mechanisms behind non-genetic resistance, because they account for cell-to-cell variability only in a static manner, by assuming randomly distributed initial protein levels. They ignore that protein levels fluctuate in single cells. Because those fluctuations can overlap with apoptotic signaling and are likely involved in the evolution of resistance in surviving cells, it appears necessary to model protein fluctuations, and not only their consequences on the initial variability. We developed a generic approach to merge in a systematic and principled manner signal transduction models with stochastic gene expression models. We describe its application to an established kinetic model of TRAIL-induced apoptosis [Spencer et al, 2009] and we show that it markedly increased model prediction capabilities: one obtains a mechanistic explanation of observations on fractional killing and non-trivial predictions of the temporal evolution of cell resistance. We discuss implications of our results, notably the predicted key role of Flip and Mcl1 and how they challenge the role of survival pathways. Boston, MA June 17-19, 2014 87 Poster Abstracts #22 : Characterization of genetic interaction classes in a metazoan reveals the modular organisation of genetic interactomes and suggests prioritizing strategies for combinatorial GWAS in human Benjamin Boucher1,2,3 , Anna Lee4,5 , Michael Hallett6,7,8 , Sarah Jenna1,2,3 1 2 3 4 5 6 7 8 Department of Chemistry, Université du Québec à Montréal, Montréal, Québec, Canada Pharmaqam, Université du Québec à Montréal, Montréal, Québec, Canada Biomed, Université du Québec à Montréal, Montréal, Québec, Canada The Donnelly Centre, University of Toronto, Ontario, Canada Department of Molecular Genetics, University of Toronto, Ontario, Canada McGill Centre for Bioinformatics, McGill University, Montréal, Québec, Canada School of Computer Science, McGill University, Montréal, Québec, Canada Rosalind and Morris Goodman Cancer Centre, McGill University, Montréal, Québec, Canada The challenge of accurately predicting genetic interactions (GIs) in human contributes to the difficulty of identifying genetic biomarkers in genome-wide association studies (GWAS). Prioritization of SNP pairs through integrative genomic approaches has the potential to overcome the computational challenge of testing every pair with combinatorial GWAS. Several in silico approaches using weighted data integration have been employed to predict GIs in the nematode Caenorhabditis elegans. Interestingly, while studies in yeast revealed the existence of a modular organization within the GI network, predictive tools targeting GIs in C. elegans still consider GIs as homogenous. To improve the performances of predictive tools for GIs, we further investigated the structure of the GI interactome in C. elegans. In this study, we identified six classes of GIs, called C1-C6. These classes are characterized by distinctive properties such as their relationship with protein-protein interaction modules, the molecular and/or biological functions of interacting genes as well as the essentiality and pleiotropy of these genes. Importantly, we identified two different classes of functional gene-modules: 1) specialized modules, a protein-module-centric class with a low pleiotropic index and 2) pleiotropic modules, a protein-module-independent class with a high pleiotropic index. Additionally, genes within these modules interact through different classes of GIs: C4 and C5 for specialized modules and C1 and C2 for pleiotropic modules. Moreover, these modules are coordinated through their interactions with “organizer” genes and also by gene-module connections found in C3 and C6. Considering this organization and the expected evolutionary constraint applied on each class of identified GIs, we propose here a strategy to prioritize SNP combinations in GWAS to more efficiently identify modifier genes/SNPs. 88 International Conference on Systems Biology of Human Disease #23 : Concise Functional Enrichment of Ranked Gene Lists Claus Kadelka1,2 , Madison Brandon3,4 , T.M. Murali5,6 1 2 3 4 5 6 Virginia Bioinformatics Institute, Virginia Tech, Blacksburg, VA Department of Mathematics, Virginia Tech, Blacksburg, VA Center for Quantitative Medicine, University of Connecticut Health Center, Farmington, CT Center for Cell Analysis and Modeling, University of Connecticut Health Center, Farmington, CT Department of Computer Science, Virginia Tech, Blacksburg, VA ICTAS Center for Systems Biology of Engineered Tissues, Virginia Tech, Blacksburg, VA Background: Advances in molecular biology have led to the broad availability of genomewide expression data as well as the development of gene interaction databases and numerous statistical algorithms to organize and extract meaning from, resp., such large data sets. Standard techniques used in genome-wide expression profiling output long lists of genes, typically ranked by differential expression level. Gene set enrichment algorithms that simultaneously analyze the expression and the interaction of all genes have been developed to reveal overall trends, previously hidden in the data. Motivation:Most gene set enrichment methods work only in a binary mode; that is, either a gene is considered differentially expressed or not. This view disregards the various levels of differential expression, which could be used to obtain a better explanation of the data. Other common shortcomings include high redundancy and non-specificity of the explanatory set, which is common among methods that analyze explanatory terms on an individual basis such as by calculating the hypergeometric p-value, which is biased toward terms that annotate many genes. To our knowledge, all gene set enrichment methods exhibit at least one of these deficiencies. Results: We present a new gene enrichment algorithm that neither disregards important information by operating in binary mode, nor shares any of the common shortcomings of other enrichment methods. Boston, MA June 17-19, 2014 89 Poster Abstracts #24 : Systems Pharmacology: Enabling early quantitative decisions in pharma from start of lead identification to clinical trials John Burke, Joshua Apgar, Andrew Sutherland Applied BioMath, LLC, Winchester, MA Systems Pharmacology modeling has been used successfully in Pharma. Here we show several case studies that have saved an estimated $295M by enabling earlier quantitative decisions. The cases include: feasibility assessments; prediction of best-in-class drug properties to enable screening, improve therapeutic window, or define clinical candidate selection criteria; and first in human dose predictions. In each case, systems approaches enabled decisions that traditional methods could not address as easily or as early in the pipeline. Unlike traditional PK/PD models, systems pharmacology models leverage known biophysical interactions and integrate data from a variety of sources (in vitro, in vivo, and clinical). Systems pharmacology models act as a central repository of data and knowledge of human pathomechanisms, allowing for the exploration of hypotheses that cannot be fully tested prior to dosing patients. These case studies show the potential of systems pharmacology approaches to enable earlier quantitative decision making with the end goal of reducing late stage attrition and delivering premier therapeutics to address unmet medical need. 90 International Conference on Systems Biology of Human Disease #25 : An Exploratory Analysis of Shared Genetic Effects between Brain Oscillations and DSM-IV Alcohol Dependence Leslie Brick1 , Valerie Knopik2,3 , Rohan Palmer2,3 1 2 3 Division of Behavioral Genetics, Department of Psychiatry at Rhode Island Hospital Division of Behavioral Genetics, Department of Psychiatry at Rhode Island Hospital Department of Psychiatry and Human Behavior, Brown University The lack of replication of genetic effects across studies of alcohol dependence (AD) has prompted increased utility of endophenotypes (EPs) that better reflect biological differences indicative of susceptibility to AD. By definition, EPs are more proximal to the underlying mechanisms of a behavior and are more heritable (i.e., h2 [> 0.70] effects). Brain oscillations are regarded as candidate EPs for AD because they may stem from regulatory genes that influence neural functioning by controlling neurochemical processes. To the extent that brain oscillations reflect differences in neuronal functioning, we explored the additive effect of common single nucleotide polymorphisms (SNPs) on the phenotypic covariance between measures of brain electrophysiology (eyes-closed resting EEG, P3 amplitude, frontal theta power, and parietal delta power) and DSM-IV alcohol dependence. Analyses were conducted using ∼1300 unrelated individuals of European descent from the Collaborative Study on the Genetics of Alcoholism. Genome-wide Complex Trait Analysis was used to estimate the SNP-heritability (h2 SNP ) of each measure and their genetic covariance with AD while correcting for ancestral background and demographics (when appropriate). h2 SNP effects varied across oscillation measures , ranging from zero to 0.85, with the greatest effects seen on theta-band oscillations. Phenotypic correlations between AD and brain oscillations were modest (<0.20). Bivariate SNP-correlations for brain oscillations were not significant (e.g., rg = 0.27 between theta-band oscillation and AD) with large standard errors. Although only preliminary and in need of replication with a much larger sample, these findings suggest that common SNPs differentially contribute to brain oscillations; however large standard errors preclude clear conclusions about associations between brain oscillations and AD in this sample. Boston, MA June 17-19, 2014 91 Poster Abstracts #26 : Sequence Analysis of the mt-COXII Gene in Crassostrea virginica: Application to Human Disease Alexis Bernstein, Alexzandrea Buscarello, Bryanna Dellaripa, Mary Disbrow, Siobhan Fennell, Ashley Heidtmann, Gabrielle Hummel, Margaret Mirabella, Alexandra Zoarski, Mary Jane Paolella Sacred Heart Academy; Hamden, CT This project targets the genome of Crassostrea virginica (Cv), the Eastern Oyster, because of the organism’s abundance, commercial value, and ecological contribution. The mitochondrial gene, COXII, (Cytochrome c Oxidase, subunit II), plays a significant role in the respiratory chain of cellular respiration. This gene has been observed in significantly higher levels in human breast carcinomas than in fibroadenomas, suggesting an association between COXII and breast cancer, of particular relevance in an all-female school. The live oysters were obtained from NOAA, National Oceanic and Atmospheric Administration, located in Milford, Connecticut. Students extracted genomic DNA from the oysters’ adductor muscles utilizing two methods: FTA card technology and Qiagen DNeasy Kits; results were verified by agarose gel electrophoresis. It was necessary to quantitate the samples, using both the Nanodrop and Biophotometer, to obtain a target concentration of 50-100ng in preparation for PCR, the Polymerase Chain Reaction. Due to the length of the gene, students designed original, overlapping primers to maximize the result. These oligos were properly resuspended to produce stock solutions and then appropriately diluted for working solutions. Students designed PCR protocols, purified the products using the spin column method and quantitated to attain a target concentration of 3-10ng for DNA sequencing . Students completed cycle sequencing using Big Dye Terminators. Their subsequent extension products were purified via an ethanol protocol which produced lyophilized pellets. In preparation for DNA Sequencing, the pellets were resuspended using 25ul of HiDi Formamide. Samples were then sequenced on the school-owned ABI Prism 310 Genetic Analyzer. The sequences were then analyzed using a variety of bioinformatics tools including NCBI, Ensembl, and UCSC Genome Browser. The results of this project were submitted to Genbank. 92 International Conference on Systems Biology of Human Disease #27 : Systemic analysis of expression data of glioblastoma multiforme patients Jaime Campos, Lars Kaderali Institute for Medical Informatics and Biometry. TU Dresden. Dresden, Germany Glioblastoma multiforme is the most aggressive malignant brain tumor. Due to its localization there is a high difficulty on its analysis and testing. It has been hypothesized that they may be different subgroups of patients, which, however, have been difficult to study due to the disease’s prognosis and characteristics. This study looks for the unique molecular relationships in different glioblastoma subgroups with the goal to find the key elements in their operation, which will allow us to classify patients and suggest promising drug targets. Expression data for 400 patients was used to classify them into 4 different subgroups with similar characteristic than those presented in literature. This information is aggregated with KEGG pathways and protein-protein interaction (PPI) networks. This integrated information was used to find enriched pathways and to study the differences in the PPI networks between subgroups. Enrichment of the KEGG pathways showed inverse activation levels in a number of pathways in 3 subgroups. The forth subgroup presented enriched pathways related to neural functions. Centrality measurements of the integrated PPI networks with the differential expression values recovered the usual suspects in cancer, such as TP53. Additionally, a number of interesting elements were found that could have a central role in each subgroup. Our analysis presents novel enrichment profiles and allows us to study the effect of different expression profiles in the context of protein interactions. In addition, it opens new questions, about the relationships between different glioblastoma patient subgroups. In future work, we will integrate additional information on molecular profiles and regulatory networks, creating a unified representation of different levels of biological regulation in glioblastoma multiforme. Boston, MA June 17-19, 2014 93 Poster Abstracts #28 : OpenBEL - A platform for capture, integration, and application of biological knowledge Natalie Catlett Selventa, Cambridge, MA OpenBEL is an open source platform for managing biological knowledge, comprised of the Biological Expression Language (BEL), a knowledgebase platform, and an application ecosystem. OpenBEL provides the ability to capture biological observations, such as molecular responses to genetic and pharmacologic perturbations, in a network format suitable for reasoning applications. BEL is a human-readable language designed to be easy for life scientists to learn and use. The BEL knowledgebase platform compiles BEL knowledge into cohesive, computable networks. BEL knowledge networks can be used for systems biology research in a broad range of biomedical research areas, for the interpretation of large-scale data sets and modeling cellular responses. Specific uses include assessment of disease and drug target biology, biomarker development, understanding drug toxicity, and evaluating translatability to model organisms. More information about OpenBEL, including documentation, example knowledge, and technical specifications can be found through the OpenBEL Portal at www.openbel.org. Support for OpenBEL is kindly provided by the OpenBEL Consortium hosted by the Linux Foundation. 94 International Conference on Systems Biology of Human Disease #29 : Discovery of kinase motifs and prediction of target substrates using the ProPeL method Michael Chou1 , Sladjana Prisic2 , Joshua Lubner3 , George Church1 , Robert Husson2 , Daniel Schwartz3 1 2 3 Department of Genetics, Harvard Medical School, Boston, MA Division of Infectious Diseases, Children’s Hospital Boston, Boston, MA Department of Physiology and Neurobiology, University of Connecticut, Storrs, CT Identification of kinase substrates is essential to understanding the role kinases play in normal and disease states. Kinases discriminate between potential substrates in part by recognizing linear sequence “motifs” of amino acids surrounding the phosphorylated residue. Current techniques for discovering kinase motifs are often expensive and labor intensive. The Proteomic Peptide Library (ProPeL) method is a novel approach to discover kinase specificity motifs de novo. Briefly, human kinases are expressed in E. coli and allowed to phosphorylate bacterial proteins. After enzymatic digestion of the bacterial lysate, tandem mass spectrometry is performed and motifs are extracted from the identified peptide sequences using the authors’ pLogo and motif-x software. The ProPeL methodology successfully recapitulated the known and diverse motifs for human basophilic Protein Kinase A (PKA) and acidophilic Casein Kinase II (CK II). We then used the motifs determined using the ProPeL methodology to directly predict with high accuracy, potential target substrates of each kinase in the human proteome using the authors’ scan-x program. We thus demonstrate the validity of the found motifs and the approach to search for potential substrates. Boston, MA June 17-19, 2014 95 Poster Abstracts #30 : Emerging landscape of oncogenic signatures across human cancers Giovanni Ciriello, Martin Miller, Bulent Arman Aksoy, Yasin Senbabaoglu, Nikolaus Schultz, Chris Sander Computational Biology Program, Memorial Sloan Ketering Cancer Center Cancer therapy is challenged by the diversity of molecular implementations of oncogenic processes and by the resulting variation in therapeutic responses. Projects such as The Cancer Genome Atlas (TCGA) provide molecular tumor maps in unprecedented detail. The interpretation of these maps remains a major challenge. Here we distilled thousands of genetic and epigenetic features altered in cancers to ∼500 selected functional events (SFEs). Using this simplified description, we derived a hierarchical classification of 3,299 TCGA tumors from 12 cancer types. The top classes are dominated by either mutations (M class) or copy number changes (C class). This distinction is clearest at the extremes of genomic instability, indicating the presence of different oncogenic processes. The full hierarchy shows functional event patterns characteristic of multiple cross-tissue groups of tumors, termed oncogenic signature classes. Targetable functional events in a tumor class are suggestive of class-specific combination therapy. These results may assist in the definition of clinical trials to match actionable oncogenic signatures with personalized therapies. 96 International Conference on Systems Biology of Human Disease #31 : Decoding Network-Attacking Mutations in Cancer Pau Creixell1 , Erwin M. Schoof1 , Craig D. Simpson1 , Nevena Zivanovic2 , Lara Perryman3 , Antonio Palmeri4 , Agata Wesolowska-Andersen5 , Hiroaki Itamochi6 , Janine Erler3 , Bernd Bodenmiller2 , Benjamin E. Turk7 , Rune Linding1 1 Cellular Signal Integration Group (C-SIG), Center for Biological Sequence Analysis (CBS), DTU, DK-2800, Lyngby, Denmark. 2 Institute of Molecular Life Sciences, University of Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland. 3 Biotech Research Innovation Centre (BRIC), Copenhagen University (KU), DK-2200 Copenhagen, Denmark. 4 Centre for Molecular Bioinformatics, Department of Biology, University of Rome Tor Vergata, 00133 Rome, Italy. 5 Functional Human Variation Group, Center for Biological Sequence Analysis (CBS), DTU, DK-2800, Lyngby, Denmark. 6 Department of Obstetrics and Gynecology, Tottori University School of Medicine, 36-1 Nishicho, Yonago 683-8504, Japan. 7 Department of Pharmacology, Yale University School of Medicine, New Haven, CT 06520, USA. Cancer cells acquire pathological phenotypes through accumulation of mutations that perturb signaling processes. While thousands of mutations have been identified, mostly by genome-wide sequencing, systematic interpretation of their role in cancer and impact on cellular information processing is presently missing. Here, we propose a computational approach (ReKINect) to identify mutations attacking signaling networks. We demonstrate six types of network-attacking mutations (NAMs) including changes in kinase modulation, network rewiring as well as the genesis and extinction of specific phosphorylation sites. Through global, quantitative analysis of the exomes and (phospho-)proteomes of five ovarian cancer cell lines we identify and validate numerous NAMs. Finally, we explore the entire cancer genome repertoire and predict hundreds of NAMs affecting kinase and SH2 driven signaling. Our approach is scalable with the complexity of cancer genomes and cell signaling, and can be readily applied in personalized precision medicine. Boston, MA June 17-19, 2014 97 Poster Abstracts #32 : Sequencing and Analysis of the mt-ATP6 Gene in Crassostrea virginica Maria Beecher, Kristina D’Agostino, Katherine Donohue, Marija Jukic, Emily Mancini, Kanita Mote, Samantha Sansone, Emily Smith, Mary Jane Paolella Sacred Heart Academy; Hamden, CT The mitochondrial ATP6 Synthase gene in the Crassostrea genus is the focus of this study. The gene is 683 base pairs in length and vital to cellular respiration’s transport chain. If mutations occur in humans, it causes neuropathies and difficulties with locomotion and vision. Recent studies concerning variations of this gene have been found to result in Leigh Syndrome, a progressive brain disorder. Mutations in human mitochondrial ATP6 are also implicated in some epithelial ovarian tumor subtypes. Crassostrea virginica, (Cv), Eastern Oysters, were provided by the National Marine Fisheries Laboratory in Milford, CT. Cv was chosen as the target organism because of its commercial relevance to Long Island Sound and its accessibility as a live organism. Invertebrate DNA was extracted from the adductor muscle utilizing two methods: FTA cards and Qiagen DNeasy Kits. Genomic DNA was then applied to low percentage agarose gels for confirmation of results. To prepare for PCR, the Polymerase Chain Reaction, DNA was quantitated using the Nanodrop and the Biophotometer to obtain the target concentration of 50-100ng. The class, who also prepared all stock and working solutions, then designed overlapping primers and the PCR protocol. Successfully amplified PCR products of the mt-ATP6 gene were later purified and quantitated for target concentrations of 3-10ng prior to DNA sequencing. The students performed all sequencing reactions and purification of all extension products to produce samples of pellet DNA. These were subsequently resuspended for utilization in the school-owned ABI Prism 310 Genetic Analyzer, a single capillary automated sequencer. Once analyzed using bioinformatics tools: NCBI, Ensembl, and UCSC Genome Browser, the sequences will be submitted for inclusion in Genbank as additions to the 24 sequences previously published by Sacred Heart Academy students. 98 International Conference on Systems Biology of Human Disease #33 : Phylogenetic profiling reveals novel functional modules in the human genome Gautam Dey, Ariel Jaimovich, Akiko Seki, Tobias Meyer Chemical and Systems Biology, Stanford University, Stanford, CA A significant fraction of human genes remain uncharacterized, posing a major challenge for both medicine and basic cell biology. While we rely heavily on phenotypic screens and high-throughput genetic interaction mapping to discover novel functions, the informationrich phylogenetic history of genes is often overlooked. We have employed phylogenetic profiles, correlated patterns of joint ortholog presence and absence across multiple genomes, to generate a map of functional interactions and predictions for hundreds of uncharacterized genes. Despite considerable success as a tool for bacterial genomics, phylogenetic profiling in eukaryotes has been hampered by a limited pool of sequenced genomes and evolutionary trends that include widespread functional divergence of duplication-derived paralogs. Taking advantage of the recent explosion in the number of fully sequenced species, we have addressed these challenges with a novel algorithm, profiling 30,000 ortholog groups across 177 eukaryotic genomes. Benchmarked using curated datasets from the literature, these profiles can be filtered into around 300 functional modules containing over 2000 genes, representing the core set of interactions that persist across species separated by millions of years of evolution. In addition, many clusters map to specific subsets of highly conserved signaling networks, highlighting the surprisingly modular architecture of pathways ranging from chromatin remodeling and splicing regulation to cholesterol biosynthesis. Importantly, the map also provides targeted, experimentally tractable predictions for over 200 uncharacterized, highly conserved genes. We have focused initially on a subset of 30 genes linked either to primary cilium formation or to a poorly characterized cluster containing the actin-nucleating WASH complex. Our experiments so far have validated the overwhelming majority of predictions, underscoring the broad utility of this resource for targeted gene function discovery. Boston, MA June 17-19, 2014 99 Poster Abstracts #34 : Integration of multi-leveled –omics data for cancer network analysis. Madeline Diekmann1,2 , Jonas Behr1,2 , Niko Beerenwinkel1,2 1 2 Department of Biosystems Science and Engineering, ETH Zurich, Basel, Switzerland SIB Swiss Institute of Bioinformatics, Basel, Switzerland Heterogeneity of cancer within and between patients poses profound challenges for data analysis and treatment. However, since all tumors need to acquire certain biological capabilities, we may expect a reduction of this variability when looking for functional implications of genetic variations. Specifically, we are interested in identifying genes that are affected similarly in a large fraction of samples. The activities of these genes may depend on various mechanisms, including genetic variations, changes in transcriptional and translational regulation, or post translational modification. High-throughput measurements of the genome, transcriptome, proteome, and phosphoproteome allow us to observe such features at an unprecedented level of detail. However, all measurements have limited sensitivity, significant technical variability and biases, and therefore illuminate the whole picture only partially. We attempt to integrate all these sources of information in a unified framework using biological networks as prior information. Accounting for uncertainty in all sources of information, we seek to identify the most likely protein activity status for all genes in a given sample. Furthermore, our model provides estimated gene activity on a continuous scale. We develop this method in the context of a mouse model for hepatocellular carcinoma, where all mentioned measurements are present. We utilize a network structure learned on public pathway databases to estimate gene activity as a combination of functions of all available data like RNA expression, protein abundance, or phosphorylation. Moreover, we also accommodate multi-leveled data on the DNA level by learning how factors such as structure, relative location, frequency and conservation of SNPs and CNVs are important for gene activity. 100 International Conference on Systems Biology of Human Disease #35 : PK-PD Modelling of the Reverse Transcriptase Inhibitor Tenofovir and Quantification of its Prophylactic Efficacy against HIV-1 Infection Sulav Duwal1 , Christof Schütte1,2 , Max von Kleist1 1 2 Department of Mathematics and Computer Science, Free University Berlin, Berlin, Germany Zuse Institute Berlin, Berlin, Germany Despite tremendous efforts, the HIV epidemics continues to expand and an effective vaccine remains to be developed. However, prophylactic strategies recently showed promise in reducing the risk of HIV onward transmission. One such strategy, pre-exposure prophylaxis (PrEP), targets uninfected persons at risk of infection. The key component of the strategy is the pro-drug tenofovir-disoproxil fumarate (TDF), which has recently been approved by the FDA for use in PrEP. However, several yet unknown factors may affect the efficacy of TDF as a part of PrEP strategies. To delineate some of these factors, we developed a pharmacokinetic model for TDF and its active anabolite tenofovir-disphosphate (TFV-DP) and coupled it with a HIV dynamics model. Our PK-PD predictions were subsequently validated with available data. Using a hybrid stochastic-deterministic simulation approach, we estimated the prophylactic efficacy against HIV-1 infections for (i) daily TDF-based PrEP, (ii) one week TDF started either shortly before, or – after viral exposure and (iii) a single dose oral TDF taken before viral challenge (sd-PrEP). The prophylactic efficacy of TDF showed a negative correlation with the number of transmitted viruses for all evaluated regimens. Once daily TDF-based PrEP with 300 mg could prevent approx. 80% infections, in agreement with clinical data (TDF-only sub-study of Partners PrEP ) and was relatively unaffected by poor adherence. However, the efficacies of event-driven prophylaxis (strategies ii-iii) were limited by a slow accumulation of TFV-DP. Finally, we derived and validated an analytic expression for the concentration vs. prophylactic effect, which summarized important parameters that affect PrEP-efficacy, such as the number of transmitted viruses etc. As TDF showed modest efficacy for the event-driven prophylaxis, in future the prophylactic efficacy of other antivirals NRTIs, NNRTIs, INIs etc. should be explored for that purpose. Boston, MA June 17-19, 2014 101 Poster Abstracts #36 : Cell-to-Cell Communication Modulates the TNF-a and IL-6 Response to TLR4 Stimulation in Macrophages: A Computational Approach Markus R. Eisele1,2 , Qiong Xue1 , Kathryn Miller-Jensen1,3 1 2 3 Department of Biomedical Engineering, Yale Universtiy, New Haven, Conneticut Institute for System Dynamics, University of Stuttgart, Stuttgart, Germany Department of Molecular, Cellular Developmental Biology, Yale University, New Haven, Conneticut Cell responses are mediated by intermediate signals that are secreted and sensed by the same cells and are subject to significant cell-to-cell heterogeneity. We recently analyzed the single-cell responses of primary human monocyte-derived macrophage (MDM) cells to lipopolysaccharide (LPS) stimulation using a high-throughput multiplexed singlecell secretomic assay, and the results showed a high variation in single-cell responses. TNF-α secretion after 20 hours was increased approximately 10-fold compared to population experiments, while in contrast IL-6 secretion was significantly lowered. We hypothesized that the propagation of intermediate secreted signals via cell-to-cell interactions might account for such differences between single cells and cells in a population. To explore this question, a computational ODE model was built using the experimental results in order to study the TNF-α and IL-6 response to LPS stimulation in single cells and small cell populations. The model reconstructs NFκB activation, followed by TNF-α secretion through LPS stimulation. TNF-α secretion can activate a second wave of cytokines, which is characterized in the model through IL-6, which can stimulate its own secretion through a strong positive feedback loop. Our simulation shows that this positive feedback is not sufficiently activated in single cells. Only cells in a population are able to activate a reliable paracrine feedback loop leading to a more homogenous response to LPS. Our results suggest that more than ten cells are needed to create a response similar to cell-population experiments. With the aid of our simulations we are able to provide an explanation regarding the signaling constraints of single cells. We suggest that isolated cells lack important paracrine feedback loops and that these feedback loops are necessary for coordinated innate immune activation following LPS stimulation. 102 International Conference on Systems Biology of Human Disease #37 : The role of p53 in metabolic regulation and response to cellular perturbations Ana Finzel Pérez, Fabian Bindel, Christin Zasada, Guido Mastrobuoni, Stefan Kempa, Alexander Loewer Berlin Institute for Medical Systems Biology, Max-Delbrueck-Center, Berlin, Germany P53 has been called “cellular gatekeeper” as it mediates the cellular response to a wide range of cellular stress signals. Upon stress, p53 is activated and stabilized by covalent modifications, showing different dynamics depending on the stimulus. It activates or represses hundreds of target genes regulating repair and cell fate. However, the correlation between p53 dynamics and temporal activation of target genes is not well known. To better understand how information transmitted by p53 is decoded, we have created fluorescent reporter cell lines for p53 and several target genes. This allows us to simultaneously monitor in individual cells p53 dynamics and target gene expression using time lapse fluorescent microscopy. We treated these cells with chemotherapeutic drugs and applied specific perturbations to identify network nodes that are critical for information decoding. Using this approach, we observed that inhibiting DNA-PK led to an increase in amplitude and width of the first p53 pulse and corresponding changes in target gene expression. As recent studies have highlighted a role for p53 in steering metabolic fluxes, we also performed pulsed Stable Isotope Resolved Metabolomics and Quantitative Proteomics to systematically study how p53 signaling affects metabolic pathways. In contrast to previous reports, we did not observe a major effect of p53 on the central carbon metabolism in unstressed conditions. We are now focusing on p53’s role in regulating metabolic changes upon genotoxic stress. By applying different chemotherapeutic drugs and specific inhibitors, we aim to identify key metabolic processes that contribute to p53’s stimulusdependent control of cell fate. By understanding how changes in p53 dynamics affect the strength and timing of target gene expression and how this in turn affects cellular regulatory networks such as metabolism, we will be able to specifically modulate the p53 response in order to change cell fate, which is important in the context of cancer therapy. Boston, MA June 17-19, 2014 103 Poster Abstracts #38 : The mechanistic role of MYCN in driving cell cycle progression in Neuroblastoma Andres Florez1 , Tatjana Ryl1 , Frank Westermann2 , Thomas Höfer1 1 2 Division of Theoretical Systems Biology (DKFZ), Heidelberg, Germany Division of Tumor Genetics (DKFZ), Heidelberg, Germany Neuroblastoma is a common extracranial tumor in children. It has been found that MYCN oncogene is an important prognosis marker in 30% of the tumors. Additional studies have found that MYCN exerts a control in the Rb-E2F1 network but the exact mechanism remains still unclear. The work by Yao et. al, has shown that E2F1 expression exhibits a bistable switch response. This translates into a controlled and irreversible commitment of cells into cell cycle upon certain threshold of growth factor stimulation. To investigate the effects of MYCN on the Rb-E2F1 switch, Neuroblastoma cell line (IMR5/75) with a Tet-inducible system for MYCN knockdown was used, and this cells line was transduced with GFP promoter construct for E2F1 and the cell cycle indicator FUCCI (Cdt1) for G1 phase determination. Steady state experiments show that cells accumulate in the G1 phase late upon MYCN knockdown. Mapping the E2F1 behavior to G1 early/late phases shows that E2F1 promoter activation is heterogeneous with cells exhibiting different E2F1 levels in G1 early and late. However the knockdown condition shows that cells failed to increase E2F1 in G1 late as compared to the control. Further qPCR validation of sorted cells by E2F1-GFP intensities confirms correlation with E2F1 mRNA levels. A mathematical model was developed to integrate these results, suggesting an important role of Rb phosphorylation in preventing cells entering into S phase. Combination of mathematical modeling and single cell experiments will lead to a better understanding of the Rb-E2F1 network and potential prediction of new drug targets based on MYCN actions. 104 International Conference on Systems Biology of Human Disease #39 : T cell immune responses generate diversity through linear cell-fate progression rather than asymmetric cell divisions Michael Flossdorf1,2 , Veit Buchholz3 , Patricia Graef3 , Dirk Busch3 , Thomas Höfer1,2 1 2 3 German Cancer Research Center, Heidelberg, Germany BioQuant Center, University of Heidelberg, Heidelberg, Germany Technische Universität München (TUM), Munich, Germany Upon infection, naive antigen-specific cytotoxic T cells expand vigorously and give rise to a population of short-lived effector and long-lived memory cells. Conflicting models have been proposed that suggest either of these subsets to be a precursor of the other or attribute their generation to asymmetrically dividing naive cells. To gain insight into the mechanism that underlies T cell diversification we combine stochastic population modeling with large scale model discrimination based on single cell in vivo fate mapping data. Our computational framework allows for stochastic differentiation and proliferation decisions of individual cells and incorporates both symmetric and asymmetric cell division. Building on this framework, we find, first, that asymmetric cell divisions of the activated naive T cells play a negligible role and, second, that phenotypic diversity is instead generated through linear cell-fate progression: Naive cytotoxic T cells give rise to slowly proliferating, long-lived subsets from which rapidly proliferating, short-lived subsets emerge. Critical predictions of this linear differentiation model have been validated in subsequent experiments. Third, we find that recall responses initiated by resting memory T cells recapitulate the primary response. We believe that our theoretical framework will be of wider applicability to the study of cell-fate decisions in proliferating cell populations. Boston, MA June 17-19, 2014 105 Poster Abstracts #40 : Human transcription factor network evolution and rewiring in disease Juan Fuxman Bass1,2 , Nidhi Sahni3,4,5 , Akihiro Mori1,2 , Shaleen Shrestha1,2 , Numana Bhat1,2 , Song Yi3,4,5 , Aurian Garcia-Gonzalez1,2 , David Hill3,4,5 , Marc Vidal3,4,5 , Albertha Walhout1,2 1 Program in Systems Biology, University of Massachusetts Medical School, Worcester, Massachusetts, USA 2 Program in Molecular Medicine, University of Massachusetts Medical School, Worcester, Massachusetts, USA 3 Center for Cancer Systems Biology (CCSB), Dana-Farber Cancer Institute, Boston, Massachusetts, USA 4 Department of Cancer Biology, Dana-Farber Cancer Institute, Boston, Massachusetts, USA 5 Department of Genetics, Harvard Medical School, Boston, Massachusetts, USA. Gene regulatory networks comprising protein-DNA interactions (PDIs) between transcription factors (TFs) and regulatory DNA regions play a critical role in development, physiology and can go awry in a variety of human diseases. In recent years, genome-wide TF occupancy profiles determined by chromatin immunoprecipitation (ChIP) greatly expanded our view of the genomic regions occupied by individual TFs in particular cell types. However, only a minority of the ∼1500 human TFs have been assayed by ChIP, and only in a limited number of cell types. As a consequence, systems-level questions relating to TF function and evolution have been difficult to address. Here we use genecentered enhanced yeast one-hybrid (eY1H) assays to interrogate the binding of more than a thousand full-length human TFs to regulatory genomic regions in parallel. Using this approach we describe the first gene-centered human enhancer network, involving 2,274 PDIs between 269 enhancers and 276 TFs. Analysis of this network reveals the evolutionary path of paralogs following gene duplication, as well as pairs of TFs that are redundant or that have opposing functions. We find that highly connected TFs are more frequently essential for viability or associated with human disease. Finally, we provide a proof-of-principle to show how eY1H assays can provide insights into disease-associated regulatory network rewiring that results from mutations in TFs or in regulatory sequences. This work greatly expands our knowledge about TF function and evolution and will provide a valuable resource to study regulatory perturbations such as those identified in genomewide association studies. 106 International Conference on Systems Biology of Human Disease #41 : Robust pattern formation in the Drosophila eye disc Avishai Gavish, Ben-Zion Shilo, Naama Barkai Department of Molecular Genetics, Weizmann Institute of Science, Rehovot 76100, Israel. cz.kebrt.html2latex.NoItemException: Can’t find specified config item cedil Periodic patterns emerge during the development of multiple tissues and organs. How those patterns are generated in a robust way is still an open question. The Drosophila eye is comprised of ∼750 eye units, named ommatidia whose crystalline order is defined during eye disc patterning through a dynamic process involving a traveling activation wave sweeping across the disc, followed by a lateral-inhibition based refinement. Using mathematical modeling, we confirm that lateral inhibition can generate patterns in the absence of spatial inhomogeneity, but find that it fails to do so in the presence of noise. We describe the basis for this breakdown and show that robustness is retrieved when a short-range diffusible activator is assumed. Experimentally, we identify this missing activator as Scabrouscedil; a previously implicated inhibitor. We further predict and verify that robustness is improved by the action of the morphogenic furrow, which effectively reduces distance between cells. We argue that mechanisms of periodic pattern formation are largely constrained by the need to buffer spatial and temporal heterogeneities. Boston, MA June 17-19, 2014 107 Poster Abstracts #42 : Multiple stimuli converge to influence transcription factor dynamics. Miriam Gutschow, Jacob Hughey, Markus Covert Department of Bioengineering, Stanford University, Stanford, CA A major challenge in understanding complex cellular signaling networks is deciphering how multiple signals can converge on one protein and be translated into a host of specific responses. Many innate immune system signaling cascades interact with transcription factor NF-κB, which in turn produces downstream effects specifically targeted to its stimuli. Our overall goal was to assess the combinatorial effects of multiple ligands converging upon NF-κB signaling. First, we stimulated cells with either Tumor Necrosis Factor (TNF) or bacterial lipopolysaccharide (LPS) at a range of concentrations spanning several orders of magnitude, and followed this with pairwise combinations of concentrations of both stimuli. We then observed the nuclear translocation dynamics of the NF-κB response in individual cells using high-throughput, live-cell imaging. Finally, we measured cytokine production in cells stimulated with representative pairwise combinations of the two ligands. We found that a greater proportion of cells were activated under the combined stimuli, however, the maximum NF-κB activation intensity did not increase when the signals overlapped. More surprisingly, we found that the effects of each individual stimulus could be used to predict the pairwise combinations as a simple linear combination. We also show through cytokine profiling, that despite the linear integration of the signaling dynamics, cells exposed to both pathogen and cytokine signals can exhibit synergistic and antagonistic downstream cytokine production, depending on the concentrations and timing of the two stimuli. 108 International Conference on Systems Biology of Human Disease #43 : Cytokine signaling in Alzheimer’s disease Levi Wood1,2 , Douglas Lauffenburger2 , Kevin Haigis1 1 2 Molecular Pathology Unit, Massachusetts General Hospital Department of Biological Engineering, Massachusetts Institute of Technology Alzheimer’s disease (AD) and related dementias are estimated to afflict more than 35 million people worldwide. The traditional approach to developing AD therapeutics has been to intervene in the formation of the amyloid beta (Aβ) plaques that are the pathologic hallmark of disease onset and progression (i.e., the amyloid hypothesis). Nevertheless, phase III clinical trials of therapies based on the amyloid hypothesis have either shown no improvement, or else have accelerated cognitive decline in patients with mild-to-moderate AD, suggesting that formation of Aβ plaques reflects only one portion of complex AD pathogenesis. Since the initial and ongoing physiologic response to Aβ involves microglial and astrocyte immune responses, we hypothesized that glial cytokines may play an important role in AD pathogenesis. Using partial least squares regression (PLSR) analysis to integrate cytokine protein expression data with pathologic disease state, we were able to identify a signature that reliably distinguished postmortem AD brain tissues from nondemented control brains (94% correctly predicted). Integration of our cytokine signaling data into a prior knowledge network of AD disease pathogenesis revealed a cluster of up-regulated cytokines in human AD tissues that may impinge upon multiple aspects of pathogenesis, including Aβ production, Tau hyper-phosphorylation, and neuronal death. Multifactorial application of these cytokines to primary neuron cultures revealed cytokines that (1) directly reduced viability and (2) reduced viability only when applied together with Aβ. These results illustrate the multi-faceted nature of AD pathogenesis and the need to study it in a multi-factorial manner in order to identify novel therapeutic targets. The cytokine signature derived from this work will serve as a basis for understanding complex signaling interactions in AD and for designing new therapeutic approaches for intervening in the disease. Boston, MA June 17-19, 2014 109 Poster Abstracts #44 : Robust DNA Repair through Collective Rate Control Tim Heinemann1,2 , Paul Verbruggen1,3 , Erik Manders3 , Gesa von Bornstaedt1,2 , Roel van Driel3 , Thomas Höfer1,2 1 Division of Theoretical Systems Biology, German Cancer Research Center (DKFZ), 69120 Heidelberg, Germany 2 BioQuant Center, 69120 Heidelberg, Germany 3 Swammerdam Institute for Life Sciences, University of Amsterdam, 1090GE Amsterdam, The Netherlands DNA repair is indispensible for the cellular protection against environmental and endogenous damaging agents. This is reflected in the increased susceptibility to cellular aging and cancer development as a consequence of impaired repair. Functional repair is carried out by enzymatic macromolecular complexes that assemble at specific sites on the chromatin fiber. How these molecular machineries and their constituent parts faithfully regulate the repair process and the repair rate in particular is poorly understood. Here we quantify nucleotide-excision DNA repair (NER) in mammalian cells and find that, despite the pathway’s molecular complexity, repair effectively obeys slow first-order kinetics. Theoretical analysis indicates that these kinetics are not due to a single rate-limiting step. Rather, first-order kinetics emerge from the interplay of rapidly and reversibly assembling repair proteins, stochastically distributing DNA lesion repair over a broad time period. Based on this mechanism, the model predicts that the repair proteins collectively control the repair rate. Exploiting natural cell-to-cell variability, we corroborate this prediction for the lesion-recognition factors XPC and XPA. Our findings provide a rationale for the emergence of slow time scales in chromatinassociated processes from fast molecular steps and suggest that collective rate control might be a widespread mode of robust regulation in DNA repair and transcription. 110 International Conference on Systems Biology of Human Disease #45 : A multiscale model of influenza A virus infection that elucidates the treatment with direct-acting antivirals Frank Stefan Heldt1 , Timo Frensing1 , Udo Reichl1,2 1 2 Max Planck Institute for Dynamics of Complex Technical Systems, Magdeburg, Germany Chair of Bioprocess Engineering, Otto-von-Guericke University, Magdeburg, Germany Influenza A viruses are human, respiratory pathogens that cause 3 to 5 million cases of severe illness and up to 500,000 deaths each year. Yet, there are currently only two classes of antiviral drugs licensed for treatment and resistant virus strains have already emerged. To identify novel targets for antiviral therapy, we constructed a multiscale model of influenza A virus infection. It accounts for the key steps of intracellular viral replication such as virus entry, viral genome and mRNA synthesis, and virus release. The model combines this information with infection dynamics on the extracellular level, i.e., the transmission of virions between host cells and the kinetics of virus-induced apoptosis. This integrated modeling approach allows us to reproduce a variety of experimental data at the single cell and the cell population level, and to predict the most promising targets for antiviral agents. In particular, we find that interference with the viral polymerase represents a very efficient treatment strategy because it interrupts the auto-catalytic mechanism of viral RNA synthesis. By contrast, targeting the steps of virus entry primarily delays virus spreading but does not protect host cells from infection in vitro. In addition, we demonstrate that cell death dynamics can strongly affect treatment success. Hence, multiscale modeling provides a systems-level understanding of viral infection and therapy, and is therefore an ideal platform to include further levels of complexity such as the immune response and between-host transmission. Boston, MA June 17-19, 2014 111 Poster Abstracts #46 : Modeling NOS3 signaling as markers for NO-bioavailability, ROS generation and cGMP-metabolism in cardiac pathophysiology Heinrich Huber1 , Michele Andreazzi1 , Melissa Swinnen1 , Peter Pokreisz1 , Stefan Janssens1,2 1 2 Department of Cardiovascular Sciences, KU Leuven, 3000 Leuven, Belgium Dep. of Cardiovascular Medicine, University Hospital Leuven, 3000 Leuven, Belgium Endothelial NO synthase (eNOS; NOS3) is key to reduce malignant reactive oxygen species and for regulating cardiovascular homeostasis by guaranteeing bioavailability of nitric oxide (NO) and cyclic GMP (cGMP). Deregulated eNOS signaling has been associated with pressure-overload-induced heart failure and dilated cardiomyopathies, and can partially result from reduced availability of its co-factor tetrahydrobiopterin (BH4) or from inhibition of its downstream soluble guanylate cyclase or pathological upregulation of phosphodiesterases (PDEs). Recent findings suggest that eNOS-related signaling eNOSmediated signaling is heavily dependent on the cell and tissue-specific molecular context and pose the need for a holistic and quantitative systems analysis. We here provide the first systems analysis of eNOS-regulated signaling. We therefore will analyze in house gene expression data of eNOS-pathway proteins in heart, brain and liver endothelium, and evaluate the tissue-specific robustness of eNOS signaling to small pathological aberrations in gene expression. Having previously identified a role of elevated phosphodiesterase 5 (PDE5) after increased cardiac afterload [1], we will analyze how differential expression patterns of PDE5 in mice and patients would translate into changes of cyclic guanosine monophosphate cGMP levels and activation of downstream protein kinase G-related signaling. We will investigate how pathological changes in PDE5 and BH4 availability may change activation of downstream NOS signaling from ultra-sensitivity (requiring a threshold for activation) to a hyper-reactive response. Finally, we will discuss how we will apply our model as potential tool for assessing the likelihood of cardiotoxic injury for patients treated with the chemotherapeutic drug doxorubicin in line with our ongoing clinical studies. 1. Vandenwijngaert S, Pokreisz P et al PLoS One, 2013. 8(3): p. e58841. 112 International Conference on Systems Biology of Human Disease #47 : Systems pharmacology analysis of drug-induced peripheral neuropathy Junguk Hur1 , Abra Guo2 , Wei-Yin Loh3 , Eva Feldman1 , Jane Bai4 1 2 3 4 Department of Neurology, University of Michigan, Ann Arbor, Michigan 48109 College of Literature, Science, and the Arts, University of Michigan, Ann Arbor, Michigan 48109 Department of Statistics, University of Wisconsin, Madison, Wisconsin 53706 Office of Clinical Pharmacology, U.S. Food and Drug Administration, Silver Spring, Maryland 20993 Background: Drug-induced peripheral neuropathy (DIPN) is a side-effect of many lifesaving drugs, often leading to dose reduction or regimen modification that may comprise clinical outcomes. The underlying mechanisms are not fully understood, and no predictive tool for DIPN is available. Here, we employed a systems pharmacology approach to identify the proteins/genes significantly associated with clinical incidence and severity of DIPN. Results: Using text-mining and expert curation, we identified 234 neuropathy-inducing drugs (NIDs) from drug labels and online resources (Drugs@FDA, DailyMed, and SIDER). A pharmacological network consisting of NIDs and their drug-targets was constructed by referencing DrugBank, and extended by adding intermediators having protein-protein or genetic interactions with the drug-targets. Comparison against 1,000 random networks identified 230 DIPN-associated intermediators, significantly enriched with apoptosis and stress-response genes. Transcriptome analysis of neural progenitor cells was performed by leveraging the Library of Integrated Network-based Cellular Signatures (LINCS) database. The transcriptomic changes caused by NIDs were highly similar to those by shRNA perturbation of the genes in the ErbB signaling pathway. We also collected neuropathy incidence and severity from drug labels and literature, and built a regression-tree model of DIPN based on the drug-targets and intermediators. Those NIDs whose targets interacted with both MYC and PAF15 were associated with a neuropathy incidence of 38.1%, while drugs interacting only with MYC had an incidence of 2.9%. Conclusions: A systems pharmacology analysis, integrating of multiple bio-/chem-informatics approaches, suggests that the ErbB signaling pathway and PAF15 may play an important role in DIPN. Boston, MA June 17-19, 2014 113 Poster Abstracts #48 : A map of functionally coherent binding for Snail and Twist transcription factors in fly mesoderm development informs regulation of epithelial remodelling and oncogenic Notch Essafi Abdelkader1 , Sims Andrew1 , Owen Jeremy2,3 , Heale Bret1 , Ford Matthew1 , Lubbock Alexander1 , Overton Ian1 1 2 3 MRC Human Genetics Unit, University of Edinburgh, Edinburgh UK Department of Systems Biology, Harvard Medical School, Boston, USA University of Cambridge, Cambridge, UK Cell identity is governed by gene expression, regulated by transcription factor binding at cis-regulatory modules (CRMs). Genome-scale assignment of functional targets is challenging and many CRMs likely remain uncharacterised. We developed and rigorously benchmarked network biology approaches to address these key limitations in understanding gene expression control and applied these novel tools to predict functional targets for the transcription factors Snail and Twist. Analysis of multiple independent datasets in fly mesoderm development found considerable unanticipated direct regulation. Predicted Snail and Twist functional targets substantially overlapped with modifiers of the Notch pathway identified by genetic, RNAi screens and also identified novel factors in processes that control Notch signalling. Notch is critical for cell fate decisions across development, linking to several cancers and genetic disorders. Snail and Twist are canonical Epithelial to Mesenchymal Transition (EMT) transcription factors, reactivation of a programme resembling EMT is a credible mechanism for key aspects of metastasis. Unsupervised clustering of microarray data for orthologues of functional Snail and Twist targets predicted by our algorithm stratified 2999 primary breast cancers by subtype. The aggressive, basal/triple-negative subtype has EMT characteristics and is driven by Notch signalling. Our integrative analysis identified conserved functional targets of EMT transcription factors in fly development as new players in basal-like breast cancers, including poorly characterised transcription factors, membrane receptors, splicing factors, vesicle trafficking proteins and chromatin modifiers. This work therefore suggests novel regulators and effectors of oncogenic Notch output relevant to the biology of cancer stem cells and metastasis. Ectopic expression of four poorly characterised predicted functional targets showed significant effects on breast cancer cell invasion. 114 International Conference on Systems Biology of Human Disease #49 : A systems biology approach to the GABAergic contribution to Juvenile Myoclonic Epilepsy Martine A Jaworski1 , Nick Tokarew2 , Yamile Wasslen3 1 2 3 Carleton University Health Services, Carleton University, Ottawa ON Canada Department of Biochemistry, Cancer Research Institute, Ottawa Hospital, Ottawa ON Canada PhD, Ottawa ON Canada Juvenile Myoclonic Epilepsy (JME) is the most common form of Genetic Generalized Epilepsy (GGE). Whole exome sequencing (WES) and other genetic studies show JME is genetically heterogeneous (eg Heinzen 2012 and OMIM EJM1-9). A JME mouse model shows abnormalities in the GABAergic system (Pal 2010; Velisek 2011); drugs enhancing GABAergic function are effective in some JME patients. Purpose: To review protein-disrupting mutations of pre- and postsynaptic GABAergic proteins in JME patients. Methods: Using systems biology, we integrated in silico data from JME studies of genetics/genomics (family studies, WES, and/or genotyping, targetted gene sequencing), as well as transcriptomics, proteomics, and synaptomics of GABAergic synapses. Results: 1. Multiple mutations in the presynaptic GABAergic vesicle life cycle occur from GABA synthesis to vesicular exocytosis and endocytosis. 2. Ionotropic and metabotropic GABAergic neuroreceptors are mutated, but mutations of their interactants in trafficking, post-translational modification, receptor desensitization, scaffolding, and endocytosis are equally numerous. 3. WES and PolyPhen-2 (Heinzen 2012) predict some deleterious mutations of AA residues at protein-protein, protein-lipid or protein-small molecule interaction sites within the GABAergic system. 4. JME mutations in GABAergic synapses target strategically placed protein complexes which modulate the Excitatory-Inhibitory (EI) balance in the motor system (eg. the putamen). Conclusions: 1. A subset of protein mutations in JME are interconnected via presynaptic and post-synaptic GABAergic pathways, converging to form a highly dynamic view of GABAergic synaptic networks. 2. Disruptive mutations in some JME patients involve proteins in GABAergic systems which directly, or indirectly, are the targets of AED, thus allowing a preliminary linkage of genomics, proteomics and therapeutics. 3. Additional knowledge of these networks at different levels of analysis may assist in understanding patient response to AED, and for patient stratification in trials of AED. Boston, MA June 17-19, 2014 115 Poster Abstracts #50 : Application of S-pyocins to eradicate Pseudomonas aeruginosa biofilms Enes Karaboga, Suphan Bakkal Sabanci University, Istanbul / Turkey Pseudomonas aeruginosa (Pa) is a gram-negative opportunistic pathogen, which is associated with life-threatening hospital-acquired and community-acquired infections occurred in urinary tract, skin, eye, ear, and lungs. Antipseudomonal antibiotics are the most potent arsenals to treat Pa infections. However, Pa infections are still among the most urgent and prevalent bacterial infections due to their elevated antibiotic resistance. Especially if Pa strains form biofilms, antibiotic resistance can be up to 1,000 times more compared to the antibiotic resistance observed in planktonic bacteria. There is an obvious need for novel therapies for the treatment of Pa infections. Recently, several studies performed on pyocins, proteinaceous bacterial toxins produced by Pa, to explore the potential of pyocins as novel antibiotics to treat Pa infections. Similar to other members of the bacteriocin family, pyocins mostly kill strains of the related species. There are three types of pyocins: S, R, and F pyocins. S-type pyocins are high molecular weight proteins and RF-type pyocins resemble bacteriophage tails. In this study, we applied four S-pyocins (S1, S2, S3, AP41) on Pa biofilms separately and in combination with six commonly used antibiotics (Tobramycin, Gentamicin, Colistin, Piperacillin, Ceftazidime, Ciprofloxacin) to determine the efficacy of S-pyocins on Pa biofilms to eradicate Pa biofilms and if there is synergy and/or antagonism between S-pyocins and antibiotics. We created Pa biofilms using Calgary Biofilm Device (CBD) and determined minimum biofilm eradication concentration (MBEC) of each antibiotics and S-pyocins separately and in combination. We expect to observe similar efficacy and also synergy between S-pyocins and antibiotics. This will indicate the potential use of S-pyocins as alternative drugs where Pa biofilms show high resistance to antibiotics. 116 International Conference on Systems Biology of Human Disease #51 : Systems Biology Evaluation of Hepatocellular Carcinoma Metabolism Antje Kettelhake1 , Guido Mastrobuoni2 , Chris Bielow2 , Ines Rudolph1 , Olga Vvedenskaya1,2 , Matthias Pietzke2 , Stefan Kempa2 , Thorsten Cramer1 1 Charité Campus Virchow-Clinic, Augustenburger Platz 1, 13353 Berlin, Germany Institute for Medical Systems Biology, Max-Delbrück-Center for Molecular Medicine, Robert-Rössle-Straße 10, 13125 Berlin 2 Hepatocellular carcinoma (HCC) is one of the most common tumor types and shows constantly rising incidence in the western world. Therapeutic options are limited since HCCs are robustly resistant to conventional treatment. Detailed understanding of the molecular HCC pathogenesis is a fundamental prerequisite to design innovative and effective forms of therapy. Deregulated cellular energetics represents an emerging hallmark of cancer. Previous work by us and others pointed towards a functional importance of glycolysis and the central carbon metabolism (CCM) for HCC malignancy in vitro and in vivo. To further analyze this, we decided to apply a systems biology approach integrating proteomic and metabolomic data from a murine HCC model system. Shotgun proteomics analyses revealed a distinct CCM pattern in murine HCC as glycolysis enzyme isoforms with high glucose affinity were found significantly upregulated. Strikingly, all CCM-regulating proteins downstream of GAPDH were downregulated compared to healthy liver. Results from metabolomics studies suggest high turn-over rates of early-glycolysis intermediates but a decelerated TCA cycle. Chemical inhibition of the glycolytic pathway pointed towards robust anti-proliferative efficacy of this approach. In summary, our data display distinct metabolic reprogramming of a murine HCC model with translation into an effective experimental therapy. Whether similar results can be obtained in human HCCs is currently being evaluated by us. Boston, MA June 17-19, 2014 117 Poster Abstracts #52 : Structure-energy-based predictions and network modelling of RASopathy and cancer missense mutations Christina Kiel1,2 , Luis Serrano1,3,4 1 EMBL/CRG Systems Biology Research Unit, Centre for Genomic Regulation (CRG), Dr. Aiguader 88, 08003 Barcelona, Spain 2 Universitat Pompeu Fabra (UPF), 08003 Barcelona, Spain 3 Universitat Pompeu Fabra (UPF), 08003 Barcelona, Spain 4 Institució Catalana de Recerca i Estudis Avançats (ICREA), Pg. Lluís Companys 23, 08010 Barcelona, Spain The Ras/MAPK syndromes (‘RASopathies’) are a class of developmental disorders caused by germline mutations in 15 genes encoding proteins of the Ras/mitogen activated protein kinase (MAPK) pathway frequently involved in cancer. Little is known about the molecular mechanisms underlying the differences in mutations of the same protein causing either cancer or RASopathies. Here we shed light on 956 RASopathy and cancer missense mutations by combining protein network data with mutational analyses based on 3D structures. Using the protein design algorithm FoldX, we predict that most of the missense mutations with destabilizing energies are in structural regions that control the activation of proteins, and only a few are predicted to compromise protein folding. We find a trend in which energy changes are higher for cancer compared to RASopathy mutations. Through network modelling we show that partly compensatory mutations in RASopathies result in only minor downstream pathway deregulation compared to cancer. In summary, we suggest that quantitative rather than qualitative network differences determine the phenotypic outcome of RASopathy compared to cancer mutations. 118 International Conference on Systems Biology of Human Disease #53 : In silico identification of biomarker-optimized treatment strategies in HER2+ cancer Daniel Kirouac, Jinyan Du, Johanna Lahdenranta, Ulrik Nielson, Charlotte McDonagh Merrimack Pharmacetuicals The use of individually tailored multi-drug combinations has been touted as a solution to improve the success of anti-cancer drug therapy, but approaches to rationally design such regimens are lacking. Herein, we combine functional proteomics and semi-mechanistic network modelling to identify strategies for treating HER2+ cancers. First, to characterize the molecular and functional diversity of the disease we performed a systematic profiling study across a panel of 20 HER2+, but otherwise heterogeneous cancer cell lines. In vitro video microscopy and LuminexTM bead-based quantitative proteomics were used to monitor cell growth and signaling responses to AKT (MK2206) and MEK (tremitinib) inhibitor treatment matrices. We find that HER2+ cell lines can be functionally classified by PI3K/AKT vs. MAPK/ERK pathway dependence, and this dependence predicted by a combination of protein measurements comprising cell surface receptors and a cell cycle regulator. PI3K and MAPK pathway suppression was found to increase the expression of multiple receptor tyrosine kinases, thereby buffering the effect of inhibitor treatments, and in some cases increasing tumor cell proliferation in subsets of cells. We next utilized a previously developed a multi-scale model of HER2+ cancer, linking cell signaling networks to proliferation and survival. Growth inhibitory responses to combinations of drugs targeting the ErbB signaling network were simulated across a diverse population of synthetic tumors. Co-targeting the HER2-HER3 dimer with a combination of Herceptin, lapatininb, and MM-111 was more effective on average than any other 3-drug combination considered. Resistance was largely confined to cells harboring activating mutations within the PI3K (PIK3CA) or MAPK (KRAS) pathways. Together, our study identified potential biomarkers, drug combinations and resistance mechanisms in HER2+ cancers, revealing strategies for optimizing the treatment of these patients. Boston, MA June 17-19, 2014 119 Poster Abstracts #54 : Inter-individual effects mask the molecular signature of psoriasis and eczema Bettina Knapp1 , Maria Quaranta2 , Natalie Garzorz3 , Martina Mattii2 , Venu Pullabhatla4 , Davide Pennino3 , Christian Andres3 , Claudia Traidl-Hoffmann3 , Andrea Cavani5 , Fabian J. Theis1,6 , Johannes Ring3 , Carsten B. Schmidt-Weber2 , Stefanie Eyerich2 , Kilian Eyerich3 1 Institute of Computational Biology, Helmholtz Center Munich, Neuherberg, Germany ZAUM - Center of Allergy and Environment, Technische Universität and Helmholtz Center Munich, DZL, Munich, Germany 3 Department of Dermatology and Allergy, Technische Universität Munich, Munich, Germany 4 Division of Genetics and Molecular Medicine, King’s College London School of Medicine, Guy’s Hospital, UK 5 Laboratory of experimental immunology, IDI-IRCCS, Rome, Italy 6 Department of Mathematics, Technische Universität Munich, Garching, Germany 2 Psoriasis and eczema are two prevalent inflammatory diseases where the underlying molecular mechanisms are not yet fully understood and the primary cause is still under debate. Due to the incomplete picture of the underlying pathogenesis, the specific therapies given for the two diseases are mainly based on empiric clinical studies rather than on targeting a specific gene or biological process. Previous attempts to use gene expression measurements to study psoriasis and eczema were limited by the high interindividual variability based e.g. upon gender and age. We present gene expression data from patients affected by both psoriasis and eczema simultaneously. Using a principal component analysis on the data we show that the inter-individual differences mask the differences between psoriasis and eczema. If we take this into account and normalize each patient’s gene expression pattern of diseased tissue against healthy tissue, we reveal intra-individual effects. Thereby, we are able to identify the molecular signatures of both diseases and to provide a comprehensive understanding of the disease pathogenesis on a single gene level but also in a broader context using a gene set enrichment analysis on the identified genes. The molecular signature of psoriasis and eczema were furthermore used to define a diagnostic tool that allows to classify the two diseases. We used an independent cohort of eczema and psoriasis patients and separated them into a training and a test set. For each patient, real-time PCR data was collected for 15 significantly regulated genes identified in the microarray analysis. Using support vector machines on the training data we established a classifier which diagnosed all patients of the test set correctly and also identified initially misdiagnosed or clinically undifferentiated patients. In future work, we will use the data not only to classify psoriasis and eczema but also to develop new therapeutic targets in a personalized manner. 120 International Conference on Systems Biology of Human Disease #55 : Identification of new IkappaBalpha complexes by an iterative experimental and mathematical modeling approach Fabian Konrath1 , Johannes Witt2 , Thomas Sauter3 , Dagmar Kulms4 1 Mathematical Modeling of Cellular Processes, Max-Delbrueck-Center for Molecular Medicine Berlin-Buch, Berlin, Germany 2 Institue for System Dynamics, University Stuttgart, Stuttgart, Germany 3 Life Sciences Research Unit, University of Luxembourg, Luxembourg, Luxembourg 4 Experimental Dermatology, TU Dresden, Dresden, Germany The transcription factor nuclear factor kappa B (NF-κB) plays a fundamental role in numerous cellular processes including apoptosis as well as inflammatory and pro-proliferative processes. Therefore, the occurrence of aberrant functionalities in the complex regulatory network of NF-κB can lead to the development of severe diseases. The NF-κB inhibitor IκBα is a crucial component of this network that sequesters the transcription factor in the cytosol by masking its nuclear localization signal. Consequently, degradation of NF-κB-bound IκBα causes the activation of NF-κB and thus transcription of its target genes. By iterative model refinement and model inspired experimentation we analysed orthovanadate-induced IκBα degradation which occurred without affecting NF-κB activity. The model’s prediction of NF-κB free IκBα complexes that contain stabilizing IKK subunits led to the design of in vitro experiments revealing the existence of a yet unknown IκBα:IKKγ complex. Together with the fact that other but canonical processes may affect the cellular IκBα status the uncovered IκBα complex might play a role in preventing undesired NF-κB activity. Boston, MA June 17-19, 2014 121 Poster Abstracts #56 : Network models of signaling and drug response in melanoma Anil Korkut, Weiqing Wang, Evan Molinelli, Arman Aksoy, Emek Demir, Chris Sander Computational Biology Center, MSKCC, New York, NY Systematic prediction of cellular response to combinatorial perturbations is a central goal in biology. To reach this goal, we developed a combined experimental-computational systems biology method, called perturbation biology. Using this strategy, we build network models of drug response in RAF inhibitor resistant melanoma cells and predict the response of the cells to previously untested drug combinations. First, we systematically perturb cancer cells with a set of targeted drugs singly and in paired combinations. Next, we measure (phospho) proteomic and phenotypic (e.g. cell cycle arrest, cellular viability) response profiles to the perturbations. High throughput (phospho)proteomic data is collected using an antibody-based proteomics platform called ZeptoMARK reverse arrays. The response profiles serve as the training set to infer network models that quantitatively link proteomic and phenotypic changes. We infer non-linear differential equation based models of cellular signaling using statistical physics algorithms. Based on the derived network models, we can rapidly predict the proteomic and phenotypic response profiles to untested drug perturbations. Experimental tests of the predictions have led to two hypotheses. (1) Inhibition of polo-like kinase (PLK1) is effective in melanoma cell lines and, at reduced concentration, may be a useful component of combination therapy in RAFiresistant melanoma. (2) Co-targeting of the c-MYC and ERK pathways has a synergistic response that may potentially overcome RAFi resistance in melanoma cells. 122 International Conference on Systems Biology of Human Disease #57 : MYCN and cellular decisions in neuroblastoma Erika Kuchen1 , Nathalie Harder2 , Tatjana Ryl1 , Chunxuan Shao1 , Andres Florez1 , Emma Bell3 , Karl Rohr2 , Frank Westermann3 , Thomas Höfer1 1 2 3 Theoretical Systems Biology, German Cancer Research Center, Heidelberg, Germany Theoretical Bioinformatics, German Cancer Research Center, Heidelberg, Germany Neuroblastoma Genomics, German Cancer Research Center, Heidelberg, Germany MYCN amplification, occurring in 20% of cases of the solid child tumour neuroblastoma, is associated with drug resistant relapse and poor prognosis. MYCN has paradoxical effects in the cell: promoting cell cycle progression and sensitising cells to apoptosis. To better understand the role of MYCN in cell fate decisions, time-lapse imaging was used to quantify cell behaviours and attributes in populations grown under a variety of conditions using MYCN-regulatable cell lines. By combining microscopy data and mathematical modelling we aim to get a mechanistic understanding of the heterogeneity in cell-fate decisions within neuroblastoma populations. Live cell imaging revealed a large heterogeneity in interdivision times and apoptotic events within both populations of MYCN overexpressing and MYCN knock-down cells. However, MYCN increased the fraction of proliferating cells and their cell cycle speed, allowing cells to evade quiescence driven by growth factor depletion. This indicates that MYCN expression level controls the switching between cycling and non-cycling states of the cell cycle. In addition, within-population variability was compared based on individual cells and their progeny. Within cell lineages sibling cells demonstrated a high similarity in their fates and attributes. This correlation deteriorated quickly in inter-generation comparisons. Taken together, our data shows that division dynamics are MYCN dependent, but retain a heterogeneity that is partly inherited. Sibling similarity suggests that divisions occur overall symmetrically, with daughter cell behaviour being partly governed by the state of the mother cell. Combining mathematical modelling and experiments, we are further investigating the nature of this control. Boston, MA June 17-19, 2014 123 Poster Abstracts #58 : TASK-2 channels contribute to pH sensitivity of retrotrapezoid nucleus chemoreceptor neurons Natasha Kumar1 , Sheng Wang1 , Najate Benamer2,3 , Sebastien Zanella4 , Yingtang Shi1 , Michelle Bevengut4 , David Penton2 , Patrice Guyenet1 , Florian Lesage3 , Christian Gestreau4 , Barhanin Jacques2 , Douglas Bayliss1 1 2 3 4 Department of Pharmacology, UVA, Charlottesville, VA, USA Universite de Nice-Sophia Antipolis, Centre National de la Recherche Scientifique (CNRS), Nice, France Laboratories of Excellence, Ion Channel Science and Therapeutics, France Aix-Marseille-Universite, CNRS, Marseille, France Central control of respiration is impaired in breathing disorders such as Congenital Central Hypoventilation Syndrome (CCHS) and central apnea. Phox2b-expressing neurons of the retrotrapezoid nucleus (RTN) function as central respiratory chemoreceptors; they are directly activated by CO2/H+, via an unidentified pH-sensitive background K+ channel, to drive breathing. Also, TASK2 (K2P5, an alkaline-activated background K+ channel) expressing RTN neurons are eliminated in a mouse model for human CCHS. Here, we test the effect of TASK2 (K2P5, an alkaline-activated background K+ channel) channel deletion on the pH sensitivity of RTN neurons and on CO2 modulation of central respiratory output. For patch clamp recordings in brainstem slices, individual RTN neurons were identified by GFP expression (driven by the Phox2b promoter) or β-galactosidase (from the gene trap used for TASK2 deletion). Whereas 95% of RTN cells from control mice were pH sensitive, only 56% from TASK2-/- mice were classified as pH sensitive; the remaining cells were pH insensitive (44%). The alkaline-activated background K+ currents were reduced in amplitude in pH-sensitive RTN neurons from TASK2-/- mice but were absent from pH-insensitive cells. TASK2 was strongly expressed in 63% of Phox2bexpressing RTN neurons (by X-gal staining), and in 85% (by a more sensitive scPCR approach). Using an in situ working heart–brainstem preparation, we found diminished inhibition of phrenic burst amplitude by alkalization in TASK2-/- mice, with apneic threshold shifted to higher pH levels. We conclude that TASK2 channels represent a molecular substrate for pH sensing in RTN respiratory chemoreceptor neurons. 124 International Conference on Systems Biology of Human Disease #59 : The Hippo Signaling Pathway Interactome Young Kwon1 , Arunachalam Vinayagam1 , Xiaoyun Sun2 , Noah Dephoure3 , Steven Gygi3 , Pengyu Hong2 , Norbert Perrimon1 1 Department of Genetics and Howard Hughes Medical Institute, Harvard Medical School, Boston, MA 02115 2 Department of Computer Science, Volen Center for Complex Systems, Brandeis University, Waltham, MA 02454 3 Department of Cell Biology, Harvard Medical School, Boston, MA 02115 The Hippo pathway controls metazoan organ growth by regulating cell proliferation and apoptosis. Many components have been identified, but our knowledge of the composition and structure of this pathway is still incomplete. Using existing pathway components as baits, we generated by Mass Spectrometry a high-confidence Drosophila Hippo proteinprotein interaction network (Hippo-PPIN) consisting of 153 proteins and 204 interactions. Depletion of 67% of the proteins by RNAi regulated the transcriptional co-activator Yorkie (Yki) either positively or negatively. We selected for further characterization a new member of the alpha-arrestin family, Leash, and show that it promotes degradation of Yki through the lysosomal pathway. Given the importance of the Hippo pathway in tumor development, the Hippo-PPIN will contribute to our understanding of this network in both normal growth and cancer. Boston, MA June 17-19, 2014 125 Poster Abstracts #60 : Functional interpretation of genome-wide association signals in arrhythmia using protein networks of cardiac ion channels Kasper Lage1,2,3 , Alicia Lundby4 , Elizabeth Rossin1,2,3 , Paul de Bakker1,2,3 , Chris NewtonCheh1,2,3 , Jesper Olsen4 1 2 3 4 Harvard Medical School Massachusetts General Hospital The Broad Institute Novo Nordisk Foundation Center for Protein Research Genome-wide association studies (GWAS) have identified thousands of loci associated with complex traits, but it is challenging to pinpoint causal genes in these loci and to exploit subtle association signals to gain biological insight. We used tissue-specific quantitative interaction proteomics to map a network of five cardiac ion channels involved in the Mendelian disorder long QT syndrome (LQTS). We integrated the ion channel network with GWAS loci from the corresponding common complex trait, QT interval variation, to identify candidate genes that were subsequently confirmed in Xenopus laevis oocytes and zebrafish. We used the ion channel network to filter weak GWAS signals by identifying single nucleotide polymorphisms (SNPs) in proximity to genes in the network supported by strong proteomic evidence. Three SNPs passing this filter reached genome-wide significance after replication genotyping in a follow up cohort of 17,500 individuals. Overall, we present a general strategy to functionally interpret GWAS loci, to integrate common and rare genetic variation involved in a particualr phenotype, and to systematically filter subtle association signals using quantitative protein interaction networks derived from the relevant tissue. 1. Lundby A, Rossin E, et al., Accepted in Nature Methods 2. Arking D, Pulit S, et al., Accepted in Nature Genetics 126 International Conference on Systems Biology of Human Disease #61 : Auto-inhibition stabilizes delayed negative feedback in biochemical systems Anastasiya Lapytsko, Jörg Schaber Institute of Experimental Internal Medicine, Otto von Guericke University, Magdeburg, Germany Negative feedback control is a fundamental and ubiquitous feature in biological systems that can serve several objectives like, e.g., stabilizing the abundance of biochemical species, inducing oscillations, modifying response times and mediating adaptation. Another feature of biological systems is time delay between a signal and its response, which appears because of the time needed to transcribe and translate biochemical information into cellular compounds. In conjunction with time delay negative feedback may lead to oscillatory behavior that might be inappropriate in cellular systems mediating adaptive responses. In order to mitigate sustained oscillations in three-dimensional signal-response systems it was suggested to couple fast and slow negative feedbacks [1-2]. We generalized this idea, where we introduced explicit time delay between signal and response that allows replacing intermediate signal transduction processes by their duration. For this we developed generic models of a cellular adaptation response using two-dimensional systems of delay differential equations and analyzed the dependence of their steady state properties on the kind of employed feedbacks. We showed that stable oscillations arise in cellular integral feedback systems due to a Hopf bifurcation. We also show that transient auto-inhibitory feedbacks can stabilize the adaptation response in terms of suppressing oscillatory behavior [3]. Thus, our general approach allows to study the nature of both oscillatory as well as adaptive behavior in cellular systems, which is demonstrated with examples from osmo-adaptation in yeast and p53 and NF-κB oscillations in mammalian cells. Our theoretical framework can be used to engineer synthetic cellular modules, in which oscillatory behavior can be fine-tuned. 1. Nguyen, L.K., (2012), J. R. Soc. Interface, 9, 1998. 2. Schaber, J., et al., (2013), J. R. Soc. Interface, 11, 20130971. 3. Lapytsko, A., et al., Submitted. Boston, MA June 17-19, 2014 127 Poster Abstracts #62 : A systems biology model for the coagulation network in non-bleeding state describes baseline activity of clotting Dooyoung Lee1 , Satyaprakash Nayak2 , Steven Martin2 , Anne Heatherington1 , Paolo Vicini3 , Fei Hua1 1 2 3 BioTherapeutic Clinical RD, Pfizer Inc, Cambridge, MA, USA Pharmacometrics, Global Clinical Pharmacology, Pfizer Inc, Cambridge, MA, USA Pharmacokinetics, Dynamics and Metabolism , Pfizer Inc, San Diego, CA, USA Coagulation is a crucial process to cease bleeding in a damaged vessel and depends on the interplay of various clotting factors. The clotting process becomes fully activated by exposing tissue factor (TF) in response to a vascular injury. Congenital deficiency of factors VIII or IX leads to hemophilia, a bleeding disorder in which patients bleed for prolonged periods after an injury. It is known that low level clotting activity exists in the normal, non-bleeding state (i.e., “baseline” conditions) as evidenced by detectable levels of prothrombin fragment 1+2 (F1+2), D-dimer and thrombin-antithrombin III complex (TAT) in healthy volunteers. However, previous coagulation models in a bleeding state have not accounted for the baseline conditions of the clotting system. The goal of this work is to understand how the baseline clotting activity impacts the coagulation process during bleeding. A non-bleeding baseline model of the human coagulation network in vivo 128 International Conference on Systems Biology of Human Disease #63 : Molecular ties between the cell cycle and differentiation in embryonic stem cells Victor Li, Marc Kirschner Harvard Medical School, Dept. of Systems Biology Attainment of the differentiated state during the final stages of somatic cell differentiation is closely tied to cell cycle progression. Much less is known about the role of the cell cycle at very early stages of embryonic development. Here we show that molecular pathways involving the cell cycle can be engineered to strongly affect embryonic stem cell differentiation at early stages in vitro. Strategies based on perturbing these pathways can shorten the rate and simplify the lineage path of ES differentiation. These results make it likely that pathways involving cell proliferation intersect at various points with pathways that regulate cell lineages in embryos and demonstrate that this knowledge can be used profitably to guide the path and effectiveness of cell differentiation of pluripotent cells. Boston, MA June 17-19, 2014 129 Poster Abstracts #64 : In Vivo Volume and Hemoglobin Dynamics of Red Blood Cells Roy Malka1,2 , Francisco Feijó Delgado3 , Scott R. Manalis3,4 , John M. Higgins1,2 1 2 3 4 Center for Systems Biology and Department of Pathology, Massachusetts General Hospital Department of Systems Biology, Harvard Medical School Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology Department of Biological Engineering, Massachusetts Institute of Technology Human red blood cells (RBCs) lose ∼30% of their volume and ∼20% of their hemoglobin (Hb) content during their ∼100-day lifespan in the bloodstream. We combine theory with single-cell characteristics to investigate the impact of vesiculation on the reduction in volume and Hb. We show that vesicle shedding alone is sufficient to explain membrane losses but not volume or Hb losses. We use dry mass measurements of human RBCs to validate the models and to propose that additional unknown mechanisms control volume and Hb reduction and are responsible for ∼90% of the observed reduction. 130 International Conference on Systems Biology of Human Disease #65 : Barcoding thousands of single cells in a single tube by droplet microfluidics Allon Klein1,2 , Linas Mazutis2,3 , Ilke Akartuna2,3 , David Weitz3 , Marc Kirschner1 1 2 3 Department of Systems Biology, Harvard Medical School, Boston, MA equal contribution School of Engineering and Applied Sciences, Harvard University, Cambridge, MA Among the central goals of single cell analysis are the identification of rare cell states, as well as cell sub-populations and their population hierarchy. For this goal, a system must be able to profile a representative sample of the target cell population, preferably consisting of hundreds or thousands of cells. There is a difficult trade-off between analyzing many cells and many individual mRNA molecules per cell. A few methods address the accessible “middle ground” of assaying up to 100-1000 cells, but this scale is not sufficient for many of the goals of single cell analysis. Here, we describe our progress in developing a technique for profiling transcriptome of at least 1,000 individual cells per single run. For this purpose, we have developed droplet microfluidic approach that delivers more than 100.000 unique DNA barcodes to individual cells in a single tube. We are using barcoded DNA primers coated on a bead that are co-encapsulated with single-cells. The advantage is that beads essentially can be used as independent reagent facilitating their handling and delivery to single-cells. Sample encapsulation, collection and processing is rapid and takes less than one hour for thousands of cells. After barcoding, the cDNA material from all cells is combined for bulk processing. The sensitivity and scale of the method is primarily limited by the depth of sequencing. Our presentation reviews technical innovations, assays of accuracy and sensitivity and our primary results of barcoding mixture of cells we have achieved to date. Boston, MA June 17-19, 2014 131 Poster Abstracts #66 : Identification of deregulated signaling pathways in Multiple Sclerosis based on gene expression data Ioannis N. Melas, Francesco Iorio, Julio Saez-Rodriguez European Molecular Biology Lab, European Bioinformatics Institute EMBL-EBI, Cambridge, UK. The construction of extensive gene expression datasets measuring cell response to perturbations is common practice these past years, in an attempt to characterize the cell’s signaling mechanisms, identify pathways that are deregulated in disease, and suggest novel drug targets. Thus, there is active research to develop methodologies to leverage these data. More specifically, one problem of interest is the use of gene expression data to identify deregulated pathways in a specific cell type. As gene expression is regulated by the activation patterns of transcription factors, and this is turn is regulated by signaling pathways, one could take advantage of changes in the gene expression level caused by a perturbations such as a drug or a disease, and backtrack all the way into the signaling level, pinpointing the proteins deregulated by that perturbation causing the observed changes in gene expression. In this work, we have developed an approach based on an Integer Linear Programming formulation for the identification of signaling networks that best fit a measured gene expression signature. In more detail, assuming we are given (i) a Prior Knowledge Network (PKN) which represents the protein connectivity according to literature, (ii) prior knowledge of transcription regulation, and (iii) a gene expression dataset capturing how a cell/tissue type of interest responds on the gene expression level; we identify subsets of the PKN and the regulon that most probably yielded the measured gene signatures. As a case study we have computed signaling models for PBMCs and identified pathways that are strongly deregulated in Multiple Sclerosis. Our findings include major players of MS as well as underreported proteins that could play a role in the disease and constitute potential drug targets. 132 International Conference on Systems Biology of Human Disease #67 : The AXL Receptor is a Sensor of Ligand Spatial Heterogeneity Aaron Meyer1,2 , Ceridwen Riley1 , Frank Gertler2,3 , Douglas Lauffenburger1,2,3 1 2 3 Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge MA Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge MA Department of Biology, Massachusetts Institute of Technology, Cambridge MA The AXL receptor is a TAM (Tyro3, AXL, MerTK) receptor tyrosine kinase (RTK) important in blood clotting, viral infection, innate immune response and cell clearance, deregulated in many human carcinomas. While the immediate cognate ligand-receptor complex (Gas6-AXL) structure is known, studies examining ligand-mediated signaling often provide paradoxical results. Therefore, a detailed, mechanistic picture of AXL activation, and thus quantitative understanding of the nature and contexts of ligand-mediated signaling, has not emerged. Employing quantitative biochemistry and deterministic modeling we show that AXL operates to sense local spatial heterogeneity in ligand concentration, due to an unusual dichotomy between ligand-dependent and ligand-independent signaling. We experimentally validate diverse model predictions concerning this behavior. The results demonstrate that AXL functions distinctly from other RTK families, and this surprising insight will be vital for the design of AXL-targeted therapeutic intervention. Boston, MA June 17-19, 2014 133 Poster Abstracts #68 : Reversible Bile Transport : A Cholesterol trader in Human Body Shekhar Mishra, Pramod R. Somvanshi, K.V. Venkatesh Department of Chemical Engineering, Indian Institute of Technology Bombay Several factors spanning from genetic to lifestyle components are responsible for defective cholesterol homeostasis leading to hypercholesterolemia. A system level analysis of the cholesterol regulatory network insights into the control mechanisms that regulate cholesterol homeostasis. A multiscale mechanistic model forms an essential basis for a control systems analysis of cholesterol processing in the human body. We have developed an integrated kinetic model to quantify the whole body cholesterol homeostasis. Using this model, we studied the effect of rate perturbations in several regulatory modules that are responsible for cholesterol homeostasis. One of the modules which exerts a significant control over cholesterol levels is the hepato-intestine bile transport system. Physiologically, bile plays a dual role, as it is responsible for consuming cholesterol for its synthesis and acts as a carrier for cholesterol absorption and transport between the organs. Thereby it controls the import and export processes of cholesterol in the human body. This, implies the necessity of a fine balance of bile salt transport in cholesterol homeostasis. It was evident from our analysis of the perturbation of hepato-intestine reversible bile transport that an increase or decrease in the rate of bile salt transport from intestine to liver or its reverse respectively, decreased total plasma cholesterol.To assess the effectiveness of these transport rates, we performed simulations in a case of induced Familial Hypercholesterolemia, which is caused by mutations in the LDLR gene. A 2fold change in the bile salt transport restored the total plasma cholesterol level from 6.5 mmol/L to 5.5 mmol/L. This suggests that the bile transport control strategy can be effectively used to enhance the cholesterol excretion flux thereby lowering the excess cholesterol levels. Thus, the bile transport module presents an alternative therapeutic method to control hypercholesterolemia. 134 International Conference on Systems Biology of Human Disease #69 : Dynamic variability of NF-kappaB signal transduction: A mechanistic model Janina Mothes, Dorothea Busse, Bente Kofahl, Jana Wolf Max Delbrück Center, Berlin, Germany NF-κB (p65/p50) is a central transcription factor in mammalian cells and drives the expression of a variety of different genes. Aberrant regulation of NF-κB is associated with chronic inflammation, asthma, neurodegenerative diseases and cancer. Therefore, NFκB is tightly regulated to ensure a correct activation of different gene expression profiles. In the absence of extracellular signals, NF-κB is inactive in the cytoplasm where it is sequestered by inhibitory κB (IκB) proteins including IκBα, IκBβ and IκB. Upon stimulation, e.g. TNFα, IκB kinases (IKKs) are activated inducing the phosphorylation and subsequent degradation of IκB proteins. Unbound NF-κB translocates into the nucleus and regulates target gene expression including genes encoding for IκB isoforms or the zinc finger protein A20. Newly synthesised IκB binds nuclear NF-κB leading to an export of the NF-κB-IκB-complex to the cytoplasm. A20 also inhibits NF-κB signalling by inhibiting pathway components upstream of IKK. This leads to an additional negative feedback. Although the dynamics of NF-κB activation is a subject of intensive research, it still remains an open question how NF-κB encodes gene regulatory information. In this context it is of special interest if the variability of the NF-κB response observed in single cells may encode different cellular outcomes. By applying mathematical modelling and bifurcation analyses, we show that NF-κB is capable of showing a great range of different dynamics, from steady states to sustained oscillations. We identified the total NF-κB concentration and the transcriptional efficiency of NF-κB on IκBα as two critical parameters that determine the NF-κB dynamics. The variation of the total NF-κB concentration might be caused by cell to cell variability. The regulation of the transcriptional effect of NF-κB on IκBα, e.g. by co-factors, might be an active way of the cell to regulate the NF-κB dynamics. Boston, MA June 17-19, 2014 135 Poster Abstracts #70 : A systems biology model of the coagulation network in bleeding state reveals differences in behaviors of biomarkers in response to perturbations. Satyaprakash Nayak1 , Dooyoung Lee2 , Sunita Patel-Hett3 , Debra D. Pittman3 , Paolo Vicini4 , Anne Heatherington2 , Steve Martin1 , Fei Hua2 1 2 3 4 Pharmacometrics, Global Clinical Pharmacology, Pfizer, Cambridge MA Biotherapeutics, Clinical Research, Pfizer, Cambridge MA Biotherapeutics RD, Pfizer, Cambridge MA Pharmacokinetics, Dynamics and Metabolism - New Biological Entity, Pfizer, LaJolla CA Blood coagulation is an important process responsible for maintaining hemostasis. The blood coagulation system is a complex and tightly regulated network of pro- and anticoagulation mechanisms which work together to initiate rapid generation of the main coagulation protein, thrombin, in response to a vascular injury or trauma and its subsequent inhibition after the blood loss has been stopped. In this work, we present a computational model of coagulation formed by combining three different models from the literature which was then numerically fit to match in-house ex vivo thrombin generation data. The model was optimized to fit 26 different Thrombin Generation Assay (TGA) experiments and activated Partial Thromboplastin Time (aPTT) in normal or hemophilic plasma simultaneously using a global optimization method. Our simulations show that biomarkers such as lag time, peak thrombin and area under the curve (AUC or Endogenous Thrombin Potential) of the TGA profile and aPTT show varying degree of sensitivity to changes in coagulation factors. For example, increasing the levels of the potent anti-coagulant anti-Thrombin III (ATIII) in the simulations changes the peak value and AUC for thrombin with little effect on the lag time. Similarly, varying the levels of active protein C (APC) had a pronounced effect on changing the TGA profile in hemophilia A plasma, but it didn’t lead to a change in aPTT. The response of endpoints depended on whether the level of an activated factor or its respective zymogen form was varied and the type of plasma, e.g., factor VIII or IX deficient plasma (simulating hemophilia A or B) that was utilized. In conclusion, our systems model predicted differences in sensitivity of biomarkers in response to various perturbations of the coagulation network. These results can serve as a useful guide in selecting the sensitive endpoints to measure in a clinical study and provide mechanistic understanding of the relative changes in biomarkers. 136 International Conference on Systems Biology of Human Disease #71 : Novel multiplex analysis of protein complexes in signaling networks reveals signatures that distinguish T-cell responses to antigen Steven C. Neier1 , Stephen E.P. Smith1 , Jason P. Sinnwell2 , Zhenjun Chen3 , Tessa R. Davis1 , Wendy K. Nevala4 , Elizabeth Ann L. Enninga4 , Scott R. Burrows5 , James McCluskey3 , Svetomir N. Markovic4 , Jeanette E. Eckel-Passow2 , Claudia Neuhauser6 , Adam G. Schrum1 1 Department of Immunology, Mayo Clinic College of Medicine, Rochester, MN Division of Biomedical Statistics and Informatics, Department of Health Sciences Research, Mayo Clinic, Rochester, MN 3 Department of Microbiology and Immunology, Institute for Infection and Immunity, University of Melbourne, Australia 4 Division of Hematology, Department of Medicine, Mayo Clinic, Rochester, MN 5 QIMR Berghofer Medical Research Institute and QIMR Berghofer Centre for Immunotherapy and Vaccine Development, Australia 6 University of Minnesota Informatics Institute, Minneapolis MN 2 As central mediators of signal transduction, protein-protein interactions (PPI) are thought to coordinate cellular responses through the formation of extensive networks. We hypothesized that measuring PPI network activity in human T cells would reveal immune activation versus tolerogenic signaling profiles. We have designed and mounted a new technological application, multiplex immunoprecipitation detected by flow cytometry (MIF), to empirically measure a significant subset of the signaling PPI network in T-cells. The approach captures protein complexes onto microspheres through immunoprecipitation, and uses fluorescent probes to detect co-associated proteins. A panel of distinct microsphere bead types (Luminex xMAPTM ) allows for capturing many distinct complexes simultaneously. Currently, MIF is capable of 253 unique pair-wise measurements between 22 different proteins that participate in T cell antigen receptor signaling. We have applied MIF to distinguish agonist (pro-immune) versus antagonist (pro-tolerance) activation of the Tcell antigen receptor. To analyze this innovative dataset, we developed new statistical approaches that utilize the full fluorescence distribution and correct for technical variability and multiple hypothesis testing. We compared our approaches to existing ones. Our results favor the use of a novel, empirically derived alpha cut-off correction, and we propose two optimized analytical approaches for MIF data. Furthermore, composite network signatures have been identified by principle component analysis, while the application of other dimensionality reduction and clustering methods continue to be explored. In summary, MIF is providing new insight into the network patterns of shared protein complexes as they translate distinct signals into T-cell function. Boston, MA June 17-19, 2014 137 Poster Abstracts #72 : A Targeted Therapy – So, what is it targeting? A Case Study of the On-Target Effect on Waldenstrom’s and the Off-Target Effect on Platelet Count and What One May Tell Us About the Other John Paasch Winsilico Chelmsford MA Deriving science from a single person case study is a challenging and some may say questionable task. While the data used for the analysis presented here is from a single person’s response to a pill that is a covalent inhibitor of Bruton’s tyrosine kinase (Btk), the person was a part of a very early and highly successful Clinical Trial (CT) of this new and novel drug. Waldenstrom’s macroglobulinemia (WM) is a rare form of non-Hodgkins Lymphoma. It is diagnosed by the presence of lymphoplasmacytic cells in the bone marrow and a monoclonal IgM found in the blood. The WM B-cells demonstrate somatic hypermutation of their antibody but do not execute Class Switch Recombination hence the excessive production of IgM. Unique to this set of data, apart from the sparse, nontime coherent samples is the degree of the dynamic response to the initial on-off protocol. This lends itself to analysis and hypothesis generation. While the on-target goal of the drug was to control the person’s WM, there was an off-target effect to the platelet count (PC). Both parameters went down when the drug was taken but rose when the drug was briefly stopped. This happened through two cycles of that protocol. The person was then switched to a continuous, lower level dose for the remainder of the CT. The focus of this treatment was addressing the intended disease. The side effects, not always of severe level, are not often analyzed and if so the results are not presented unless they are critical or life threatening. I will report the analysis followed to take this data and extrapolate from related lab work reported by others to create a hypothesis of what have been happening, PBPK, to create the undesired off-target, PD, effect to the PC. The intent of this effort is to show that it may be possible to further the understanding of the intended mechanism of action of this drug by using the results seen in other cell types within the blood cell group and identify areas that could be explored further. 138 International Conference on Systems Biology of Human Disease #73 : Mathematical modelling of the regulation and degradation of the transcription factor EB Marin Zapata Paula A.1 , Jünge Anja1 , Brady Nathan R.2 , Hamacher-Brady Anne1 1 2 Lysosomal Systems Biology, German Cancer Research Center (DKFZ) and Bioquant Systems Biology of Cell Death Mechanisms, German Cancer Research Center (DKFZ) and Bioquant Autophagy is a catabolic process in which cytosolic components are engulfed by doublemembrane vesicles, and subsequently degraded after fusion with lysosomes. At the transcriptional level autophagy is largely regulated by the recently characterized transcription factor EB (TFEB), which translocates to nucleus in response to metabolic and lysosomal stresses. In this work we combine computational and experimental approaches to study the regulation of TFEB, focusing on spatial and temporal degradation events. By integrating literature knowledge with sequence similarity from related transcription factors, we developed different competing ODE-models for the TFEB regulatory network, which satisfy reported findings. These models were compared against each other and against our time course Western blot data of endogenous TFEB. Obtained results allowed us to discard some of the models, since they were unable to reproduce the experimental data. Importantly, some of the remaining models implemented novel regulatory links based on sequence similarity, and thus suggest a narrowed experimental validation strategy. Taken together this work exemplifies how mathematical modeling can drive experimental hypothesis testing, and reduce focus to a set of possible regulatory models. Boston, MA June 17-19, 2014 139 Poster Abstracts #74 : Metabolomic characterization and analysis of liverspecific functions of an in vitro fibrosis model Christian Priesnitz1 , Roman Liebe2 , Yeda Kaminski1 , Frank Lammert3 , Fozia Noor1 1 Biochemical Engineering Institute, Saarland University, Saarbruecken, Germany Molecular hepatology and alcohol associated diseases, Medical faculty, Heidelberg University, Mannheim, Germany 3 University Hospital of Saarland University, Homburg, Germany 2 Fibrosis is characterized by an excessive formation and deposition of fibrotic connective tissue and occurs as a result of inflammation or tissue damage in the liver. Factors leading to the development of fibrosis are among others TGF-ßwhich is secreted by various cell types in the liver upon tissue damage. TGF-ßis known to be a major pro-fibrotic cytokine that induces epithelial-mesenchymal transition (EMT) of hepatocytes in vitro. Furthermore, it plays critical roles in the onset and progression of liver fibrosis in vivo. Most of the research in this field is still conducted using animal models. We compared two mouse strains, namely C57/BL6 and DBA/2J which are commonly used in the field of fibrosis research. We applied a metabolomics approach since the metabolome represents the integrated phenotype of regulation on the genome, transcriptome and proteome level. We analyzed hepatocytes of both strains regarding their central carbon metabolism and liver-specific functions, such as urea and albumin production and CYP450 activity, under control conditions as well as in the presence of TGF-ß. The analysis of the viable cell numbers and LDH release over time as well as the liver specific functions show clear strain-dependent differences in the susceptibility of hepatocytes in vitro to TGF-ß. We observed a significant decrease in the viable cell number within 5 days in culture and an earlier onset of dedifferentiation in DBA/2J hepatocytes as compared to C57/BL6 hepatocytes. However, we did not detect any TGF-ßspecific signature in the central carbon metabolism. This shows that the central carbon metabolism in contrast to most liver specific functions is robust against TGF-ßexposure. In combination with proteome and transcriptome data, these results provide a valuable basis for strain selection for studies focusing on the systematic in-depth understanding of fibrosis and fibrosis related diseases as well as research focusing on detection of early biomarkers in the fibrosis related disease 140 International Conference on Systems Biology of Human Disease #75 : Network topological characteristics aid in identifying causal genes for atherosclerosis Holly Arnold1,2 , Stephen Ramsey1,2 1 2 Department of Biomedical Sciences, Oregon State University, Corvallis, OR School of Electrical Engineering and Computer Science, Oregon State University, Corvallis, OR A central problem in bioinformatics is leveraging molecular profiling data, in conjunction with prior knowledge (e.g., the global protein interaction network) to predict key molecules that initiate or regulate pathogenesis. Understanding the molecular basis of pathogenesis of atherosclerosis is both vital for uncovering new therapies and an appealing test-case for investigating how lists of molecules with known roles in the disease (e.g., from mouse gene-targeting studies and from rare disease-causing mutations) can be used to guide a computational strategy for predicting disease genes. We hypothesized that the location and connectedness of a protein within the global protein interaction network (i.e., the local and global network topology) can be used, in conjunction with molecular profiling data from plaque, to predict genes that are involved in atherosclerosis in the context of a machine-learning algorithm. To test this hypothesis we identified 222 genes with probable causal roles in atherosclerosis, from the primary research literature, the NCBI OMIM database, and the JAX MGI database, as a training set of molecules (“positive cases”). As features for the machine learning, we used eight topological characteristics extracted from the protein interaction network as well as data from transcriptome profiling studies of plaque. Using the Random Forest machine-learning algorithm and a ten-fold cross-validation strategy, we investigated the prediction performance (area under the full sensitivity-vs.-FPR curve) of the algorithm for different subsets of features and the relative benefit of each feature for prediction performance. We found that as a predictive feature, shortest-paths proximity to a positive case (a known disease gene) is highly beneficial for accurately predicting disease genes and that it strongly outperforms all the other features. Our results indicate the importance and the potential of network-based approaches for prioritizing candidate disease genes. Boston, MA June 17-19, 2014 141 Poster Abstracts #76 : The Effect of Higher-Order Receptor Clusters on TRAIL Induced Apoptotic Signaling Raue Andreas1,2 , Chai Diana1 , Hudson Hannah1 , Tam Eric1 , Schoeberl Birgit1 1 2 Merrimack Pharmaceuticals, Cambridge, MA, USA University of Freiburg, Freiburg, Germany The trimeric TNF-related apoptosis-inducing ligand (TRAIL) is an endogenous ligand that binds to trimeric death receptors (DR4/5). TRAIL is known to induce apoptosis mainly in malignant cells while normal cells remain unaffected. This makes TRAIL a most interesting target for cancer therapy. We investigate the effects of ligand valency on the degree of Caspase-8 activation and apoptosis induction. As a model system we use the semi-sensitive cell line DU145 and multivalent anti-DR5 fibronectin domains that can potentially form higher-order receptor clusters. In order to disentangle the effects of receptor clusters of different size, a dynamic model of ligand-receptor binding and Caspase-8 activation is coupled to a model of cell growth and cell death. The dynamic model is calibrated using quantitative experimental data of ligand on cell binding, Caspase-8 activity and cell viability. Our results show that higher-order receptor clusters do amplify apoptotic signaling. The model predicts that receptor clusters of size three and six are more abundant than others, indicating the special role of preformed receptor trimers. However, Caspase-8 activation only increases after receptor clusters of size larger than three. On the cell population level, the model can quantitatively predict the induction of apoptosis in this cell line. This work is an important step towards an integrative model of TRAIL induced apoptosis that finally aims at the understanding of heterogeneity both on a single cell level across cell lines. 142 International Conference on Systems Biology of Human Disease #77 : Organ-specific stochastic phenotype switching is required for endothelial health Lei Yuan1,2 , Gary C. Chan1,2 , David Beeler1,2 , Lauren Janes1,2 , Katherine C. Spokes1,2 , Anahita Mojiri3 , William J. Adams4 , Tracey Sciuto4 , Guillermo Garcia-Cardeña5 , Grietje Molema6 , Nadia Jahroudi3 , Philip A. Marsden7 , Ann Dvorak4 , Erzsébet Ravasz Regan1,2 , William C. Aird1,2 1 Center for Vascular Biology Research, Beth Israel Deaconess Medical Center Harvard Medical School, Boston MA 3 Department of Medicine, University of Alberta, Edmonton, Alberta, Canada 4 Department of Pathology, Beth Israel Deaconess Medical Center, Boston, MA 5 Center for Excellence in Vascular Biology, Department of Pathology, Brigham and Women’s Hospital, Boston, MA 6 Department of Pathology and Medical Biology, Medical Biology Section, University Medical Center Groningen 7 Department of Medical Biophysics, University of Toronto, Ontario, Canada 2 Among unicellular organisms, stochastic phenotype switching is a documented strategy for survival. These populations hedge their bets: while the majority of their cells are adapted to their present environment, a minority remains poised to thrive under drastically different conditions. Bet hedging has also been described in metazoan cells, primarily in vitro. However, its role in tissue homeostasis has yet to be established. Here we show that von Willebrand factor (vWF) is expressed in a spatially heterogeneous manner in a small fraction of capillary endothelial cells in the heart, skeletal muscle, lung and brain. To test whether this fraction is dynamically maintained, we generated transgenic mice that allow for direct comparison between snapshots of vWF mRNA expression (vWF+/LacZ ) and its cumulative expression over time (vWF+/Cre crossed with a LacZ reporter line, leading to permanent LacZ expression in transiently vWF+ cells). We have found that vWF mosaic patterns are dynamic, in that vWF expression stochastically toggles ON/OFF over time. By contrast, expression in the aorta and liver is static in time. Mosaic vWF heterogeneity is also present in cultured primary endothelial cells from multiple vascular beds, and it is stochastically reconstituted in clonally expanded populations. We show that the ON/OFF bimodality of vWF expression is due to an epigenetic switch, the ON/OFF states of which require low/high DNA methylation of the vWF promoter. Stochastic transitions between active/silenced vWF transcription were accompanied by dynamically changing DNA methylation, resulting in its heterogeneity among clonal cells. Finally, vWF-/- mice demonstrated extensive endothelial cell damage in capillaries of the heart, but not kidney or aorta. Taken together, these findings suggest that dynamic mosaism of vWF expression is functionally relevant and represents represent a novel, adaptive homeostatic Boston, MA June 17-19, 2014 143 Poster Abstracts mechanism tailored to specific organs. 144 International Conference on Systems Biology of Human Disease #78 : Treating monogenic human diseases with evolving computationally designed mutagenic triplex-forming and recombinagenic donor DNA molecules Faisal Reza, Peter M. Glazer Yale University, School of Medicine, New Haven, Connecticut 06520, U.S.A. Endogenous human genome targeting and editing in an efficient and specific manner are technological challenges, particularly in development and translational settings, with significant foreseeable impacts on healthcare. The use of synthetic molecules to confer safe and effective therapeutic changes to the human genome is sought in translational medicine. Targeting and editing one or more genes in an efficient and specific manner remains a challenge in addressing monogenic human diseases. To address these challenges, we have developed and applied designed triplex-forming molecules, composed of synthetic oligo- or peptide- nucleic acids, to target and elevate the DNA repair and recombination machinery in human progenitor cells by forming triplex structures within duplex chromosomal and episomal DNA. Once elevated, we have developed and applied designed recombinagenic donor DNA molecules as homologydependent repair templates to co-opt this DNA repair and recombination machinery to introduce edits into the genome. Safety and efficacy is achieved by leveraging the performance profile of the cell’s own endogenous DNA repair machinery in concert with these sequence-specific and localizingin-tandem molecules. Progenitor cells drugged with these designed molecules for triplexformation and for recombination, and primed with chemical cell modulators, safely and effectively target and edit the genome, which are then propagated to cellular progeny, for physiological gene correction, induction, and regulation. This triplex-forming and recombinagenic molecular technology is developed to remediate the underlying genetic causes of human diseases, such as hemoglobinopathies, and has a well-positioned technology profile and potential for human systems and synthetic biology as well as translational medicine. Boston, MA June 17-19, 2014 145 Poster Abstracts #79 : Topological Measures of Protein-Protein Interactions And Implications For Cancer Therapy Sebastien Benzekry1 , Edward Rietman2 , Jack Tuszynski3 , Giannoula Klement2 1 2 3 Institut de Mathematiques de Bordeaus, Bordeaux, France Newman-Lakka Institute, Tufts University School of Medicine, Boston, MA 02111 Department of Oncology, University of Alberta, Edmonton, Alberta, Canada T6G 1Z2 A correlation between the complexity of cancer protein-protein interaction networks and 5-year survival of cancer patients has been shown (Breitkreutz et al., 2012; Takemoto et al., 2013). Here we extend that work to include two related topological measures. We first show a linear correlation of a topological measure with 5-yr survival and then we show that inhibition of specific proteins in the network, as suggested interpolation on the survival curve, may actually lead to improved survival. This new approach may prove applicable to selecting drug targets for cancer therapy. 1. Breitkreutz D, Hlatky L, Rietman E, Tuszynski JA. Molecular signaling network complexity is correlated with cancer patient survivability. Proc Nat Acad U S A. 2012 Jun 5;109(23):9209-12. 2. Takemoto K, Kihara K. Modular organization of cancer signaling networks is associated with patient survivability. Biosystems 2013 113(3):149-154. 146 International Conference on Systems Biology of Human Disease #80 : Identification of novel CD2AP SH3 interaction partners by combining crystallography, bioinformatic tools and peptide arrays Jenny Rouka1 , Philip C. Simister1 , Melanie Janning1,2 , Joerg Kumbrink3 , Tassos Konstantinou1 , Tobias Kroger4 , Joao R.C. Muniz4 , Moin Saleem5 , Stefan Knapp4 , Frank Von Delft4 , Nicola O’Reilly6 , Stephen Taylor7 , Rudolf Volkmer8 , Kathrin Kirsch3 , Stephan M. Feller1 1 2 3 4 5 6 7 8 Biological Systems Architecture Group, Department of Oncology, WIMM, University of Oxford, Oxford, UK UKE Hamburg, Department of Oncology/Hematology, Hamburg, Germany Boston University Medical College, Department of Biochemistry, Boston, MA, USA Structural Genomics Consortium, Oxford, UK Bristol Heart Institute, University of Bristol, Bristol, UK Cancer Research UK Peptide Synthesis Group, London Research Institute, London, UK Computational Biology Research Group, University of Oxford, Oxford, UK Charité, Institute of Medical Immunology, Berlin, Germany CD2AP is a member of the CD2AP/CIN85 family adaptors and involved in several cellular processes, such as kidney podocyte development and actin-mediated membrane trafficking. It may also be involved in tumorogenesis. It contains three SH3 domains whose binding properties and interaction partners remain largely unexplored. The CD2AP SH3 interaction with the novel partner Rab5-activating GEF RIN3 was studied extensively by isothermal titration calorimetry (ITC), peptide scanning arrays, mutagenesis studies and X-ray crystallography. Mapping of the interaction regions showed that human RIN3 contains two binding sites for the CD2AP SH3 domains. From these studies, the CD2AP SH3 preferred binding motif P-x-P/A-x-x-R emerged. Two crystal structures (1.6 Åand 1.2 Å) of the SH3-1 and SH3-2 domains in complex with RIN3 epitope I and II respectively revealed that these residues serve as anchoring points. With the aid of bioinformatics tools, this motif was used to conduct a peptide array based screen for additional signalling partner candidates. One of the hits was the Arf-GAP ARAP1. ITC data indicate that the three SH3 domains differentially recognise three ARAP1 epitopes, with the first ARAP1 epitope binding to SH3-2 in the nanomolar range. A crystal structure (1.6 Å) of the SH3-2 domain in complex with the first ARAP1 epitope uncovered two additional anchoring residues that extend beyond the PPII helical region of the binding epitope. The CD2AP - ARAP1 interaction was confirmed in podocytes and cancer cells at the endogenous protein level. Even though RIN3 and ARAP1 are also involved in membrane trafficking, a direct link to CD2AP had not been reported before. Further candidates from the peptide array analyses were also investigated by ITC. Boston, MA June 17-19, 2014 147 Poster Abstracts In conclusion, this study led to the elucidation of the CD2AP SH3 domain binding signatures and the identification of putative novel binding partners. The intriguing CD2AP ARAP1 interaction warrants further investigations. 148 International Conference on Systems Biology of Human Disease #81 : Characterizing heterogeneity in leukemic cells using single cell gene expression analysis Assieh Saadatpour1,2 , Guoji Guo1,3 , Stuart Orkin1,4,5 , Guo-Cheng Yuan1,2 1 2 3 4 5 Dana-Farber Cancer Institute Harvard School of Public Health Boston Children’s Hospital Boston Children’s Hospital Howard Hughes Medical Institute Characterization of cancer heterogeneity is of immense importance with significant clinical implications. Recent advances in single-cell gene expression analysis have paved the way to identify rare cell types and to dissect potential intra-cancer cellular hierarchy.Here, we apply two complementary approaches to characterize the heterogeneity within leukemic cells at the single-cell resolution using the MLL-AF9 driven mouse model of acute myeloid leukemia (AML). We started with fluorescence-activated cell sorting (FACS) analysis with seven surface markers, and extended our study by using a multiplexing quantitative polymerase chain reaction (qPCR) approach to assay the transcriptional profile of a panel of 175 carefully selected genes in leukemic cells at single cell level. We employed an integrative set of computational tools to identify the cellular hierarchy within leukemic cells and to investigate transcriptional network changes. Our analysis showed striking heterogeneity among the leukemic cells. Mapping to the hematopoietic cellular hierarchy identified two distinct subtypes of leukemic cells; one similar to granulocyte/monocyte progenitor (GMP) phenotype and the other to macrophage and dendritic cells (DCs).Network analysis further revealed similarities as well as organizational differences between the GMP and leukemia networks. Overall, the combined analysis presented in this study pinpoints previously unrecognized heterogeneity within leukemic cells and provides insights into the molecular signatures of AML. Boston, MA June 17-19, 2014 149 Poster Abstracts #82 : Characterization of the protein signatures of permissive vs. non-permissive influenza A virus infections in human host cells by quantitative proteomic analysis Anne Sadewasser1 , Katharina Paki1 , Katrin Eichelbaum2 , Matthias Selbach2 , Thorsten Wolff1 1 2 Div. of Influenza viruses and other Respiratory Viruses, Robert Koch-Institute, Seestr. 10, 13353 Berlin Max-Delbrück-Center for Molecular Medicine, Robert-Roessle-Str. 10, 13125 Berlin Influenza virus infections are the major cause for respiratory disease in humans, which affect all age groups and result in extensive global mortality and morbidity, as well as substantial economic costs. Human and most avian influenza A virus (IAV) strains differ largely in their replication efficiency in and activation of human cells despite successful cell entry. We hypothesize that the distinct outcome of an infection with a given virus strain is determined by the differential interplay between specific host and viral factors, which remain to be defined in their entity. By mass spectrometry-based quantitative proteomics we aim to characterize the sets of cellular and viral factors whose abundance is specifically up- or down-regulated in permissive and non-permissive IAV infection, respectively. Our analysis involves a “Spike-in SILAC (stable isotopic labeling by amino acids in cell culture)” approach in human A549 cells that are highly permissive for a seasonal H3N2 strain, but restrict replication of an avian H3N2 virus. This allows the description and comparison of changes in the host cell proteomes in response to infections with these two viruses. The analysis resulted in the identification of a distinct set of cellular factors influenced by viral infection. Most identified cellular and viral proteins were regulated in a similar way for both virus strains, but also candidates with distinct changes in permissive and non-permissive infection were found. Ongoing studies involve the validation of the detected proteins and further bioinformatic and functional analyses to elucidate their roles in IAV infection and their potential contribution to the species barrier. In conclusion, we will present initial results of a comprehensive analysis on the complex host pathogen interaction network by using the efficient tools and methods of system biology. 150 International Conference on Systems Biology of Human Disease #83 : Training of signaling pathways to phosphoproteomic data via hybrid node/edge optimization using a mixed integer programming formulation Theodore Sakellaropoulos, Leonidas G Alexopoulos National Technical University of Athens Biological networks have always been an integral part of Systems Biology’s effort to model mathematically the biological processes of the cell. Despite the vast amount of identified interactions stored in databases, modeling the cell behavior is not a straightforward task since it is heavily dependent on the biological content in which the cell resides. In this work, we tried to couple the prior knowledge available in the pathway databases with experimental data describing the response of epithelial cells to certain stimuli perturbations in order to create a cell-specific network depicting the signal transduction mechanism of the cell. To this end, we utilized the data available for the recent SBV Improver Species Translation Challenge. In particular, starting from the reference network provided for the challenge we augmented it with reactions derived from literature to represent our prior knowledge about the connectivity of the proteins of interest. Signal transduction in the network was modeled using Boolean logic as reactions represented Boolean gates between nodes. We then formulated an mixed integer linear program that sought to modify this network by removing nodes and reactions in order to better describe the experimental data at hand. Each modification was associated to a cost. Edges that appear in multiple databases as well as nodes measured in the dataset required stronger evidence to be removed. Finally, extra constrains were added to the problem to guarantee that the final topology would be a directed acyclic graph, a structure that would allow us to infer causal relationships among the involved compounds. Boston, MA June 17-19, 2014 151 Poster Abstracts #84 : Study on the Clinical Significance of c-FLIP and JUN-B Expression in Psoriatic Lesion by using Quantitative RT-PCR and Tissue Microarray MN Salleh1 , HI Lai1 , N Shamsudin2 , WK Wan Ramli1 , MI Mustafa1 1 2 Faculty of Biomedical and Health Sciences, University of Selangor, Dermatology Unit, Department of Medicine, Faculty of Medicine and Health Sciences, University Putra Psoriasis is a chronic T cell-mediated inflammatory skin disease. Studies have shown that angiogenesis plays an important role in the pathogenesis of psoriasis. In the present study, we investigated the involvement of c-FLIP and Jun-B protein expression in the possible mechanisms of psoriasis. The hyperproliferation of keratinocytes observed in psoriasis prompted us to evaluate c-FLIP and Jun-B expression on biopsies collected from involved and uninvolved skin of 42 samples with active plaque-type psoriasis with respect to healthy skin. We analyzed the expression of C-FLIP and Jun-B at transcript and protein levels by quantitative RT-PCR and tissue microarray based immunohistochemistr. We demonstrated the expression of both genes c-FLIP and Jun-B in a hyperproliferative skin condition not related to neoplastic transformation. Interestingly, we observed that c-FLIP and Jun-B is not expressed in healthy skin, but it becomes detectable in non-lesional areas and it is even more expressed in lesional psoriatic skin. In addition, we found that Jun-B expression is correlated to the rate of keratinocyte proliferation and activation compared to c-FLIP. Hence, our observations indicate Jun-B as a new possible player, involved in the development and/or maintenance of the hyperproliferative state of psoriatic keratinocytes. Our study confirms distinct prognostic relevance of c-FLIP and Jun-B protein expression levels in the psoriasis patients and identifies an association of high Jun-B levels with elevated expression of Jun-B target genes and markers for infiltrating immune cells. 152 International Conference on Systems Biology of Human Disease #85 : Pulsatile FoxO3a translocation in response to growth factors is controlled jointly by AKT and ERK pathways Somponnat Sampattavanich1 , Bernhard Steiert1,2,3 , Bernhard Kramer1,4 , John Albeck5 , Peter Sorger1 1 Department of Systems Biology, Harvard Medical School, Boston, MA Institute of Physics, University of Freiburg, Germany 3 Freiburg Center for Systems Biology, University of Freiburg, Germany 4 Division of Systems Biology of Signal Transduction, German Cancer Research Center, Heidelberg, Germany 5 College of Biological Sciences, University of California, Davis 2 FOXO factors serve as important gate keepers of cell growth although it is unclear how mitogenic signaling along the AKT and ERK signaling cascades get integrated by FOXO proteins. Using a panel of RTK ligands that exhibit differing activation levels of ERK and AKT pathways, we demonstrated here that both AKT and ERK signaling cascades uniquely contribute to the FoxO3a translocation dynamics. We first showed that the relative FoxO3a phosphorylation at ERK-specific versus AKT-specific sites (S294/S253) correspond well with the relative ERK/AKT phosphorylation activity by the different RTK ligands. Using both immunostaining and live cell imaging, we then showed that AKTdominating ligands promote persistent cytosolic localization of FoxO3a whereas ERKdominating ligands promote transient nuclear-to-cytosolic translocation of FoxO3a initially, and later exhibit pulsatile FoxO3a translocation. Therapeutic inhibition of MEK phosphorylation does not affect the early non-stationary FoxO3a translocation but reduce the late pulsatile response. Investigation of the FoxO3a/AKT/ERK relationship in different breast cancer cell lines reveals alternative role of ERK pathway on FoxO3a dynamics, by suppressing the activation of AKT on FoxO3a nuclear-to-cytosolic translocation. Collectively, our study reveals the sophisticated dynamics of FoxO3a translocation by different RTK ligands and emphasizes the role of FoxO3a as an important integrative node of both AKT and ERK signaling cascades. Boston, MA June 17-19, 2014 153 Poster Abstracts #86 : Speech abnormalities in Parkinson’s disease reflect a major disturbance in sensorimotor servomechanisms Shimon Sapir Department of Communication Disorders, University of Haifa, Haifa, Israel Parkinson’s disease (PD) is a progressive and highly debilitating disease of the central nervous system, affecting the basal ganglia and their cortical and subcortical networks. Nearly all individuals with PD have voice and speech abnormalities, collectively termed hypokinetic dysarthria (HKD). The HKD is typically characterized by a weak and hoarse voice, reduced prosodic intonation, reduced articulatory movements, manifested as imprecise production of vowels, consonants, and other phonemes, and a tendency to accelerate (festinate), thus resulting in mumbled speech (Sapir et al., 2007). Until recently, these speech abnormalities have been attributed to dopamine deficiency and muscle rigidity. However, research and clinical studies indicate that HKD is a comlex and dynamic disorder, involving multiple factors, most of these are related to deficits in sensorimotor servomechanisms (Sapir, 2014). This poster will present two patients with PD and HKD to illustrate the speech abnormalities and the effects of behavioral treatment designed to improve motor speech control through optimization of servomechanisms and vocal vigilance. 1. Sapir, S. (2014, in press). Multiple factors are involved in the dysarthria associated with Parkinson’s disease: A review with implications for clinical practice and research. Journal of Speech Language and Hearing Research. 2. Sapir, S., Spielman, J., Ramig, L., Story, B., & Fox, C. (2007). Effects of intensive voice treatment (the Lee Silverman Voice Treatmen (LSVT)) on vowel articulation in dysarthric individuals with idiopathic Parkinson’s disease: acoustic and perceptual findings. Journal of Speech, Language, and Hearing Research, 50, 899-912. 154 International Conference on Systems Biology of Human Disease #87 : Systems Biology Approach to Determine the Efficacy of Methylation by DNA-Methyl Transferase1 in Cancer Cells Eric Samorodnitsky, Emily Ghosh, Sahana Majumder, Sibaji Sarkar Cancer Center, Department of Medicine, Boston University School of Medicine, Boston, MA Epigenetic regulation of gene silencing involves both histone deacetylation-methylation and DNA CpG island methylation. Maintenance of DNA methylation is mediated by DNA methyl transferase 1 (DNMT1). DNMT1 expression level varies with stages of cell cycle in normal cells but it is highly expressed in cancer cells. DNMT1 methylates the new DNA strand after replication, at the sites complimentary to the hemimethylated mother strand to maintain a constant methylated status when and after the cell division. Promoter region methylation silences many tumor suppressor genes in cancer cells. We have shown that re-expression of tumor suppressor genes by using histone deacetylase inhibitors could be exploited to develop novel combination therapies (Anti Cancer Research, 2012). We proposed how epigenetics could be regulating cancer progenitor cell formation and metastasis (Epigenomics 2013, 2014; IJMS 2013, Anti Cancer Research 2014). However, it is not known how DNMT1 maintains higher methylation level in cancer cells. We employed a modified Hill equation to determine the cooperativity of binding (n) of DNMT1 to the hemimethylated upstream CpG sites in cancer cells compared to the normal cells. Increase in “n” value determined the affinity of DNMT1 towards the site of methylation. We observed that the value of “n” for DNMT1 in cancer cells compared to the normal cells is higher, even when we consider enhanced expression of DNMT1 in cancer cells. Our results suggest that DNMT1 efficiency is higher in cancer cells to maintain enhanced methylation of tumor suppressor genes. This is the first demonstration of application of systems biology approach to epigenetics to explain gene silencing. Our system presents a wide applicability in calculating and determining the genome wide efficacy of methylation and gene silencing. Boston, MA June 17-19, 2014 155 Poster Abstracts #88 : Comprehensive identification of cooperative microRNA target regulation through an integrative workflow Ulf Schmitz1 , Shailendra K Gupta1 , Xin Lai2 , Felix Winter1,2 , Julio Vera2 , Olaf Wolkenhauer1,3 1 Department of Systems Biology and Bioinformatics, University of Rostock, 18051 Rostock, Germany Department of Dermatology, Faculty of Medicine, University of Erlangen-Nuremberg, 91054 Erlangen, Germany 3 STIAS, Wallenberg Research Centre at Stellenbosch University, Stellenbosch, South Africa 2 MicroRNAs (miRNAs) regulate gene expression at the post-transcriptional level in most cell-biological processes. Deregulated miRNA expression can lead to or mediate the emergence and progression of human diseases. There is increasing evidence that pairs of miRNAs can cooperatively repress translation of mutual target genes. In our own work we have shown that this phenomenon can realize fine-tuned and adaptive target gene regulation. However, up to now it was not clear which miRNA pairs may cooperate and what are their targets. We here describe a computational workflow to identify and analyze targets of synergistic miRNA regulation. This workflow integrates bioinformatics and systems biology approaches in six steps that iteratively increase the level of confidence on predicted triplexes composed of two miRNAs and their mutual target mRNA: (i) identification of miRNA binding sites in the 30 untranslated region of gene targets; (ii) identification of putatively cooperating miRNA pairs with binding sites in close proximity; (iii) prediction and analysis of the local secondary structure of putative RNA triplexes; (iv) prediction of the 3D triplex structure including Argonaute and the determination of the triplex’ thermodynamic profile by molecular dynamics simulation; (v) determination of binding affinities among the involved molecules; and (vi) determination of the target repression efficiency by simulations of a kinetic model. In a case study, we have analyzed the human genome to identify targets of synergistic miRNA-mediated repression. Our analysis provides evidence that the phenomenon of cooperative target regulation by miRNA pairs is a common cellular mechanism in human. A functional enrichment analysis that we performed on the identified targets indicates relevance of miRNA-cooperativity in cancer (signaling-) pathways. 156 International Conference on Systems Biology of Human Disease #89 : Diseaseome 2.0: Uncovering disease-disease relationships through the human interactome Amitabh Sharma1,2 , Joerg Menche3 , Albert-László Barabási1,2,3 1 Channing Division of Network Medicine, Harvard Medical School, Boston, MA-USA CCNR, Dept. Of Physics, Northeastern University, Boston-MA-USA 3 Department of Theoretical Physics, Budapest University of Technology and Economics, H1111, Budapest, Hungary 2 Human diseases could be viewed as perturbations of networks. A thorough understanding of the complex topological and dynamical properties of the interactome (Human protein interaction network) is therefore crucial to explain the mechanisms of such perturbed biological systems. There is extensive evidence in the literature that disease genes are not placed at random in the interactome. Indeed, they have a higher chance of interacting with each other than expected by chance indicating that these genes tend to cluster in the same region of the network. According to the ‘disease module hypothesis’ the cellular components associated with the same disease segregate in the same neighborhood of the human interactome. Identifying this network neighborhood is not only essential to understand the molecular mechanisms responsible for the disease, but could help predict novel drug targets and biomarkers. Yet, given the known incompleteness of the interactome and our limited knowledge of disease-associated genes, the available data may not yet have sufficient coverage to allow us to map out the modules associated with each disease. Here, we derive the mathematical conditions for the identifiability of disease modules, revealing a list of diseases for which the current interactome is sufficient to uncover the corresponding disease module. We show that the network-based location of each disease module determines its pathobiological relationship to other diseases, finding for example that diseases with overlapping network modules show significant co-expression patterns, symptom similarity, and comorbidity. In contrast, diseases residing in separated network neighborhoods are clinically distinct. The developed tools allow us to build an interactome-based platform to predict the common molecular roots of such clinically unrelated diseases as asthma and celiac disease, or to identify the biological role of low significance genes identified by genome-wide association studies. Boston, MA June 17-19, 2014 157 Poster Abstracts #90 : Proximity of metabolic and cancer causing genes in the genome leads to metabolic remodeling in cancers. Ashwini Kumar Sharma1 , Roland Eils1,2 , Rainer König1,3 1 Division of Theoretical Bioinformatics, German Cancer Research Center (DKFZ), Heidelberg. Institute of Pharmacy and Molecular Biotechnology, Bioquant Center, University of Heidelberg, Heidelberg. 3 Integrated Research and Treatment Center, Center for Sepsis Control and Care (CSCC), Jena University Hospital, Jena. 2 Cancer metabolism – an emerging hallmark of cancer, is pivotal to cancer progression and survival. Crucial metabolic genes (MG) have been identified as oncogenes (OG) and tumor suppressors (TSG) or targets of deregulated signaling. Cancer is a direct consequence of genomic alterations, somatic copy number alterations (SCNA) that frequently disrupt the genome, occurs across cancer types targeting multiple genes at the affected loci. However, the underlying challenge is to delineate the driver genes from the functionally neutral ones. The present work aims to identify how spatial proximity in the positioning of metabolic and cancer causing genes (CG) in the genome leads to metabolic remodeling. Our focus was restricted to MGs, where we observe that they are often co-altered i.e. co-deleted or co-amplified, with key CGs across cancer types, being proximal and sharing the SCNA susceptible loci. We hypothesize that there could be a subset of such co-alteration events where the genomic proximity driven alteration of the MG could further have a functional impact. A hybrid method based on statistical tests, correlation and network analysis was developed to identify such driving co-alterations events. Briefly, gene pair distances (within a chromosomal arm) is computed for all protein coding genes and their copy number co-alteration frequencies calculated across 16 cancer types using publically available cancer genomics data generated by the TCGA project. The method then identifies for each cancer type, those CG-MG pairs whose copy number co-alteration frequencies are significantly different from other gene pairs within the same chromosomal arm and separated by similar gene distances. Furthermore the CG-MG pairs are prioritized by integrating expression data, correlating copy number changes to gene expression. Finally, we identify functionally important CG-MG pairs by filtering for those, where neighboring genes of the MG in a metabolic network also show differential pattern of expression. 158 International Conference on Systems Biology of Human Disease #91 : Harnessing DNA Damage Repair Pathways in Breast Cancer Therapy: a Synthetic Lethality Paradigm Sriganesh Srihari Institute for Molecular Bioscience, The University of Queensland, St. Lucia, QLD 4072, Australia With an estimated 1.38 million new cases and 458000 deaths worldwide every year, breast cancer is the most common malignancy among women worldwide. A major contributor to these deaths is a subset of breast tumours that exhibit high relapse and metastasis rates and do not respond well to traditional therapies. Modulating DNA-damage response (DDR) pathways has shown immense potential as a specialized therapy to counter aggressive tumours, by inducing high levels of DNA damage in cancer cells thereby forcing them into apoptosis while minimizing the impact on normal cells. Exploiting synthetic lethal (SL) interactions has shown some promising results in this regard, most notably in BRCA1-deficient cells that are sensitive to PARP inhibition. A SL relationship between two genes exists when cells remain viable when either or both genes are active, but selective killing of cancer cells occurs when both genes are inactivated. However, except the BRCA1-PARP1 breakthrough few other new SL targets have successfully proceeded to clinical trials and been adopted in the treatment.Here, we seek to identify novel SL relationships among components of the DDR machinery that can be effectively translated to treatment of aggressive breast tumours. Through extensive literature searching, we have curated six DDR pathways to generate a comprehensive and up-to-date map of genes, reactions and mechanisms underlying the DDR machinery. An extensive evaluation of these pathways against known databases such as KEGG and Reactome shows we had added at least 101 new genes in 68 new reactions. Furthermore, combining graph-theoretic modelling integrating protein-interaction and gene-expression datasets with comparative genomic approaches for extrapolating SL relationships from lower-order eukaryotes such as yeast, we have identified a subset of gene-targets in these pathways which are now in the pipeline for siRNA-mediated depletion and validation in the lab.We foresee that our collaborative computational and experimental efforts focused towards identification of SL relationships will eventually lead to discovery of specialized drug targets for treating aggressive breast tumours. Boston, MA June 17-19, 2014 159 Poster Abstracts #92 : Network modeling reveals key features of epithelialto-mesenchymal transition dynamics Steven N. Steinway1,2 , Jorge Gomez Tejeda Zañudo3 , David J. Feith2 , Thomas P. Loughran, Jr.2 , Reka Albert3 1 2 3 Penn State Hershey Cancer Institute, The Pennsylvania State University University of Virginia Cancer Center, University of Virginia Department of Physics, The Pennsylvania State University Epithelial-to-mesenchymal transition (EMT) is a developmental process hijacked by cancer cells to leave the primary tumor site, invade surrounding tissue, and establish distant metastases. A hallmark of EMT is the loss of E-cadherin expression, and one major signal for the induction of EMT is transforming growth factor beta (TGFβ), which is dysregulated in up to 40% of hepatocellular carcinoma (HCC). We have constructed an EMT network of 69 nodes and 134 edges by integrating the signaling pathways involved in developmental EMT and known dysregulations in invasive HCC. We then used discrete dynamic modeling to understand the dynamics of the EMT network driven by TGFβ. Our network model recapitulates known dysregulations during the induction of EMT and predicts the activation of the Wnt and Sonic hedgehog (SHH) signaling pathways during this process. We show, across multiple murine (P2E and P2M) and human HCC cell lines (Huh7, PLC/PRF/5, HLE, and HLF), that the TGFβ signaling axis is a conserved driver of mesenchymal phenotype HCC and confirm that Wnt and SHH signaling are induced by TGFβ in mesenchymal but not epithelial cell lines. Furthermore, we identify by network analysis nine regulatory feedback motifs that stabilize the EMT process and show that these motifs involve cross-talk among multiple major pathways. We employ a node perturbation screen to identify interventions that would abolish epithelial-to-mesenchymal transition. We demonstrate that the TGFβ driven EMT is robust to suppression of individual nodes and that 40 out of 2,346 possible combinations of node knockouts suppress EMT. Furthermore, we demonstrate that the knockout combinations act by disrupting feedback loops that regulate the EMT process. These results establish network modeling as an important tool to identify critical mediators in complex biological processes. We further propose network modeling as a tool to discover therapeutic targeting strategies within complex disease pathways, specifically in liver cancer invasion. 160 International Conference on Systems Biology of Human Disease #93 : Ras mutant-specific response to upstream inhibition Edward Stites Department of Pathology and Immunology, Washington University School of Medicine, Saint Louis, MO EGFR inhibitors are a recommended treatment for patients with colorectal cancer patients. Patients with an activating KRAS mutation are generally resistant to EGFR inhibitors, with the exception of patients that have the activating KRAS G13D mutation. KRAS G13D shares the same general biochemical defects as the other oncogenic KRAS mutants, and a mechanism to explain the differential response remains to be determined. The effects of Ras mutants on upstream (EGFR) inhibition were here studied with a computational model of Ras signaling that has proven useful for studying oncogenic Ras mutants. Simulations find an increased response to upstream inhibition for Ras G13D mutants compared to G12D and G12V mutants. As simulations utilize the known biochemical properties of these mutants, the simulations reveal that differential sensitivity to upstream inhibition is indeed consistent with known Ras biology. Further analysis identifies a single, readily measurable property to be the parameter that indicates response to EGFR inhibition. Measurement of this property for the >90 KRAS mutants that have to date been observed in colon cancer may provide a method for evaluating the likelihood of whether a patient with a KRAS mutation might benefit from treatment with an EGFR inhibitor. Of note, two KRAS mutants other than G13D have been measured to have this property. More generally, this work demonstrates how mass-action modeling of oncogene containing networks can contribute to the advancement of personalized medicine. Boston, MA June 17-19, 2014 161 Poster Abstracts #94 : Delineating G-protein coupled receptor (GPCR)-linked oscillatory signaling mechanisms through single cell microfluidic analysis and computational modeling Madhuresh Sumit1,2 , Richard Neubig3 , Jennifer Linderman1,2,4 , Shuichi Takayama1,4,5 1 2 3 4 5 Biophysics Graduate Program, University of Michigan, Ann Arbor, MI 48109 Department of Chemical Engineering, University of Michigan, Ann Arbor, MI 48109 Department of Pharmacology and Toxicology, Michigan State University, East Lansing, MI 48824 Department of Biomedical Engineering, University of Michigan, Ann Arbor, MI 48109 Department of Macromolecular Science and Engineering, University of Michigan, Ann Arbor, MI 48109 Although GPCRs are the target of around 40% of all pharmaceuticals, the signaling pathways by which these drugs work are not yet fully understood, raising questions about their off-target effects and limiting applications. Our work addresses this problem by developing a quantitative understanding of selective modulation of GPCR-linked signaling. It combines pulsatile microfluidic experiments (providing precise temporal control), and computational modeling (ensuring that the model developed is verifiable and can be used to predict the effects of drugs). Our current study focusses on the muscarinic receptor M3, which is implicated in type 2 diabetes and multiple sclerosis. Using a novel microfluidic platform to deliver low concentration, pulsatile ligand stimulation of GPCRs, we quantify calcium signaling and NFAT translocation in single HEK293 cells expressing M3 receptors through fluorescence imaging of R-GECO1 (calcium-sensor) and NFAT4-eGFP. Temporally patterned ligand stimulations with varying concentration (C), duration (D), and rest period between pulses (R) are delivered to modulate the cell signaling, and in silico model is predicted based on the microfluidic analysis. Our results show, for the first time, that GPCR-based calcium oscillations (triggered by nano-molar pulses of ligand) lead to efficient and sustained transcriptional activation as compared to a non-oscillatory stimulation. We have also quantified the relationship between amplitude and frequency down-regulation with receptor density, which provides insights into receptor regulation during dynamic signal processing. These findings will be used to study cells with low receptor densities (mimicking diseased states), and to develop a quantitative understanding of the effects of various modulating agents (e.g., PAMs, NAMs, and node inhibitors) for targeted drug discovery. 162 International Conference on Systems Biology of Human Disease #95 : A new stochastic kinetics approach to quantitative description of intracellular networks coupled to hidden cell environment Jaeyoung Sung1 , Kim Ji-Hyun1 , Lim Yu Rim1 , Park Seong Jun1 , Yang Gil-Suk1 , Song Sanggeun1 , Yang Sora2 , Lee Nam Ki2 1 2 Department of Chemistry, Chung-Ang University, Seoul, Korea Department of Physics, POSTECH, Pohang, Korea We present a new stochastic kinetics approach to a quantitative description of intracellular networks coupled to hidden cell environment in which the product creation rate coefficient and the reactant level are unknown stochastic variables that vary from cell to cell and fluctuate over time. A general relationship of the product number fluctuation to the creation rate fluctuation and the product lifetime is established for a basic unit of intracellular networks. An application of our approach to the central dogma of gene expression provides a unified quantitative explanation about various gene expression statistics. Our analyses provide an accurate quantification of the intrinsic and extrinsic noises, which reveal the general structure of the intrinsic and extrinsic noises in terms of the promoter strength. This work suggests a new direction in quantitative analyses of the chemical fluctuation produced by intracellular networks coupled to hidden cell environment about which we don’t have enough information to construct an explicit model. Boston, MA June 17-19, 2014 163 Poster Abstracts #96 : Global metabolic interaction network of the human gut microbiota Jaeyun Sung1 , SungHo Jang2 , Seunghyeon Kim1,3 , Nicholas Chia4 , Yong-Su Jin5,6 , Gyoo Yeol Jung2 , Pan-Jun Kim1,3 1 Asia Pacific Center for Theoretical Physics, Pohang, Gyeongbuk, South Korea Department of Chemical Engineering, Pohang University of Science and Technology, Pohang, Gyeongbuk, South Korea 3 Department of Physics, Pohang University of Science and Technology, Pohang, Gyeongbuk, South Korea 4 Department of Surgery, Mayo Clinic, Rochester, MN, USA 5 Department of Food Science and Human Nutrition, University of Illinois at Urbana-Champaign, Urbana, IL, USA 6 Institute for Genomic Biology, University of Illinois at Urbana-Champaign, Urbana, IL, USA 2 Background: Despite recent advances in knowledge of the microbial diversity inside the human gastrointestinal tract, the global interaction dynamics between the myriad microbial species, and its influence on host health and disease remains poorly understood. To better understand the emergent properties arising from the interactions between human gut microbes, we constructed a global network model based on interspecies cross-feeding relationships. Methods and Results: We used transport reaction information from public databases (e.g. KEGG, BioCyc, CAZy, TransportDB), from published genome-scale metabolic models, and from literature annotations to identify small-molecule metabolic compounds that are imported and/or exported by microbes found to reside in the human gut. Next, we defined an interaction between two microbes when one class of species can uptake a metabolic compound that is secreted by another (i.e. interspecies cross-feeding). Following this approach, we linked all interacting microbes into a global Microbe-Microbe Network. Using microbiome samples collected from patients across four clinical phenotypes, we superimposed each sample’s microbial abundance information upon our Microbe-Microbe Network, thereby removing low-quantity species (i.e. nodes), along with their interactions (i.e. edges). This led to interaction networks specific to individuals, and, in turn, networks specific to phenotype. Interestingly, we identified network-based topological features, as well as enrichment of biologically meaningful interspecies interactions, unique to each clinical phenotype. Conclusion: We present the first microbial community structure within the human gastrointestinal tract based on interspecies cross-feeding of small-molecule metabolites. Network analysis of the global microbial symbiosis provides novel insight into the molecular basis 164 International Conference on Systems Biology of Human Disease of pathophysiology. Furthermore, our results can be utilized for clinical applications in the form of network-based disease classifiers. Boston, MA June 17-19, 2014 165 Poster Abstracts #97 : Hepatic Response to Refeeding - From Ketone Bodies to Lipids Christian Tokarski1 , Sebastian Vlaic2 , Reinhard Guthke2 , Stefan Schuster1 1 Department of Bioinformatics, Friedrich Schiller University of Jena, Germany Leibniz Institute for Natural Product Research and Infection Biology e.V. - Hans-Knöll-Institute (HKI), Jena, Germany 2 Despite several important functions of the liver including detoxification of drugs, alcohol, ammonia and other xenobiotics, as well as synthesis of compounds like bile, membrane lipids and proteins, it is in particularly important in energy storage after meals and in providing energy during periods of starvation. In times of low blood glucose level the brain, which cannot metabolize fatty acids, is generating energy additionally to glucose from ketone bodies that are solely produced in the liver. In the liver these are synthesized by degradation of fatty acids via β-oxidation and exported into the blood. Since a high level of ketone bodies in blood for prolonged periods leads to damage of the nervous system, kidney and to a lower blood pH value, it needs to be lowered. Ketone bodies are predominantly utilized in skeletal muscle, heart and brain. The liver itself is only partially able to utilize them since the two important enzymes limiting the ketolysis, succinyl CoA-oxoacid transferase (SCOT) and methylacetoacetyl CoA thiolase (MAT), show very low activity in liver tissue. In order to study the role of the liver in ketone body metabolism, we created a literature based minimal model of the human fatty acid metabolism. Published gene expression data from Vollmers et al. was downloaded from GEO (acc. no. GSE13093). Since this data captures the molecular changes in response to re-feeding mice after a period of starvation, we seeked to observe differential gene expression among the genes included in our minimal model. To investigate the relations in the expression of genes we performed network inference using ExTILAR. Including prior knowledge of diverse types in the inference we obtained a dynamic gene regulatory network that is able to reproduce the observed dynamics. Connecting this network with the minimal model of hepatic fatty acid metabolism, the whole response of the liver to refeeding can be simulated leading to incorporation of ketone bodies into the liver which seem to be metabolized to form cholesterol or fatty acids via high activity of acetoacetyl-CoA synthetase (AACS) and acetyl-CoA acetyltransferase 2 (ACAT2). 166 International Conference on Systems Biology of Human Disease #98 : Reconstruction and in silico analysis of metabolism in the apicomplexan parasite Toxoplasma gondii Stepan Tymoshenko1,2 , Rebecca Oppenheim3 , Rasmus Agren4 , Jens Nielsen4 , Dominique Soldati-Favre3 , Vassily Hatzimanikatis1,2 1 Laboratory of Computational Systems Biotechnology, École Polytechnique Fédérale de Lausanne, CH1015 Lausanne 2 Swiss Institute of Bioinformatics, CH-1015 Lausanne, Switzerland 3 Department of Microbiology and Molecular Medicine, University of Geneva, CMU, 1 Rue Michel Servet, CH-1211 Geneva 4 Department of Chemical and Biological Engineering, Kemivägen 10, Chalmers University of Technology, SE412 96 Gothenburg Toxoplasma gondii is one of the most wide-spread human pathogens worldwide. Chronic infection with this eukaryotic parasite is asymptomatic and generally not harmful for immunocompetent individuals. Yet in the case of an acquired immunodeficiency or an immunosuppressive therapy reactivation of toxoplasmosis often causes an acute and lifethreatening disease. Current options for treatment of toxoplasmosis are limited and not well-tolerated as well as inefficent against the encysted, slow growing form. There is a clearly unmet need for new medication and metabolism is a promising source of drug targets. Over the last decade in silico metabolic modelling has been extensively exerted for studying genotype-phenotype relations and prediction of phenotypes upon various perturbations. Metabolic models provide an efficient framework for exploring potential vulnerabilities in metabolism of pathogenic species and thus facilitate discovery of novel drug targets. ToxoNet1 is a genome-scale metabolic model of T. gondii built in silico using state-of-the-art automated reconstruction algorithm (RAVEN toolbox). The reconstruction process and subsequent flux-balance analysis of the model allowed us to get a systematic overview of the metabolic capabilities of this pathogen. In particular, the model identified a number of gaps in the current knowledge of Toxoplasma metabolic pathways and clarified its minimal nutritional requirements for asexual replication. Using functionality of RAVEN toolbox called metabolic tasks we further defined the set of alternative precursors salvage of which is necessary for the parasite. Within simulated human host cell environment ToxoNet1 predicts a minimal set of 53 enzyme-coding genes to be indispensable for parasite replication. Some of these predictions represent readily testable hypotheses that can facilitate identification of novel potential targets and their combination for an effective intervention against toxoplasmosis. Boston, MA June 17-19, 2014 167 Poster Abstracts #99 : MOCAT: a metagenomics assembly and gene prediction toolkit Jens Roat Kultima, Peer Bork European Molecular Biology Laboratory MOCAT is a highly configurable, modular pipeline for fast, standardized processing of single or paired-end sequencing data generated by the Illumina platform. The pipeline uses state-of-the-art programs to quality control, map, and assemble reads from metagenomic samples sequenced at a depth of several billion base pairs, and predict proteincoding genes on assembled metagenomes. Mapping against reference databases allows for read extraction or removal, as well as abundance calculations. By mapping reads to a database of single-copy phylogenetic marker genes it is possible to calculate the abundance of metagenomic operational taxonomic units (mOTUs). MOCAT runs on UNIX machines and integrates seamlessly with the SGE, PBS and LSF queuing systems, commonly used to process large datasets. Individual processing steps and parameters were benchmarked and tested on artificial, real, and simulated metagenomes resulting in an improvement of selected quality metrics. MOCAT can be freely downloaded at http://www.bork.embl.de/mocat/. 168 International Conference on Systems Biology of Human Disease #100 : Controllability of protein interaction network identifies human disease genes Arunachalam Vinayagam1 , Yang-Yu Liu2,3,4 , Bahar Yilmazel5,6 , Ho-Joon Lee7 , Charles Roesel5,6 , Yanhui Hu1,5 , Young Kwon1 , Amitabh Sharma2,3,4 , Norbert Perrimon1,8 , AlbertLászló Barabási2,3,4 1 Department of Genetics, Harvard Medical School, Boston, MA Center for Complex Network Research and Department of Physics, Northeastern University, Boston, MA 3 Channing Division of Network Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 4 Center for Cancer Systems Biology, Dana-Farber Cancer Institute, Boston, MA 5 Drosophila RNAi Screening Center, Department of Genetics, Harvard Medical School, Boston, MA 6 Bioinformatics program, Northeastern University, Boston, MA 7 Department of Systems Biology, Harvard Medical School, Boston, MA 8 Howard Hughes Medical Institute, 77 Avenue Louis Pasteur, Boston, MA 2 The protein-protein interaction (PPI) network is crucial for cellular information processing and decision making. With suitable inputs, PPI networks drive the cells to diverse functional states, such as cell proliferation or cell death. Here, we characterized the controllability of a large directed human PPI network comprised of 6,339 proteins and 34,813 interactions, allowing us to classify the proteins as “indispensable”, “neutral” and “dispensable” with respect to network controllability. We find that 21% of the proteins are indispensable from a control perspective. Interestingly, these proteins are the primary targets of disease-causing mutations, human viruses, and drugs, suggesting that altering a network’s control principles is critical for the transition between healthy and disease states. Further, analyzing copy number alterations data from 1,547 cancer patients reveals that 56 genes that are frequently amplified or deleted in nine different cancers are “indispensable” from control perspective. As 46 of them have not been previously associated with cancer, it suggests that network controllability will be useful to identify novel disease genes and potential drug targets. Boston, MA June 17-19, 2014 169 Poster Abstracts #101 : Assessment of the human gut microbiome in response to temporal and colorectal cancer-associated variability Anita Voigt1,2,3 , Georg Zeller1 , Julien Tap1,4 , Shinichi Sunagawa1 , Jens Roat Kultima1 , Paul Costea1 , Petra Schrotz-King5 , Matthias Kloor2,3 , Cornelia M. Ulrich5 , Magnus von Knebel Doeberitz2,3 , Iradj Sobhani4 , Peer Bork1,3 1 Structural and Computational Biology Unit, European Molecular Biology Laboratory Heidelberg, Germany Dept. of Applied Tumor Biology, Institute of Pathology, University Hospital Heidelberg, Germany 3 Molecular Medicine Partnership Unit, University of Heidelberg and European Molecular Biology Laboratory, Germany 4 Department of Gastroenterology and LIC-EA4393-EC2M3, APHP and UPEC Université Paris-Est Créteil, Créteil, France 5 Division of Preventive Oncology, National Center for Tumor Diseases (NCT) Heidelberg, Heidelberg, Germany 2 The gut is a habitat of microbial communities (microbiota) that are very important for health e.g. by stimulating the immune system and changes of the microbiota have been linked to diseases such as obesity, type II diabetes and also colorectal cancer (CRC). Before drawing conclusions about disease-associated microbiota changes we assessed the temporal variability of the gut microbiome in seven healthy subjects using metagenomic shotgun sequencing. We found that the inter-individual variability of the undisturbed microbiota is high and although it undergoes fluctuations over time, the microbial signatures are highly individual-specific at species-level. However, perturbations like antibiotics can lead to a pervasive change of the gut microbial composition. We further investigated changes in the gut microbiota associated with CRC as several bacterial species (like Fusobacterium nucleatum) have been implicated in the development of CRC. We tested the potential of the microbiota to separate CRC patients from controls and this microbesbased test was slightly better than the standard screening fecal occult blood test and could be validated in independent patient and control populations from different countries. 170 International Conference on Systems Biology of Human Disease #102 : High-Throughput Proteomic and Transcriptomic Data Integration based on MS/MS and RNA-Seq Data using Prior Pathway Knowledge Astrid Wachter1 , Thomas Oellerich2 , Jasmin Corso3 , Ekkehard Schütz4 , Annalen Bleckmann1,5 , Henning Urlaub3,6 , Tim Beissbarth1 1 Department of Medical Statistics, University Medical Center Göttingen, Göttingen, Germany Department of Hematology/Oncology, Johann Wolfgang Goethe University Frankfurt, Frankfurt, Germany 3 Max Planck Institute for Biophysical Chemistry, Bioanalytical Mass Spectrometry Group, Göttingen, Germany 4 Chronix Biomedical, 5941 Optical Court Suite 203E, San Jose, CA 95138, USA 5 Department of Hematology/Oncology, University Medical Center Göttingen, Göttingen, Germany 6 Bioanalytics, University Medical Center Göttingen, Göttingen, Germany 2 Nowadays information acquired by different high-throughput technologies allows greatly accelerated research turning data integration into a very current topic. Although a lot of studies deal with integration of high-throughput data sets, not much emphasis is placed on incorporating prior biological knowledge into data integration methods yet. We implemented a three-level (protein, transcription factor and transcript/gene level) pathwaybased integration approach which benefits from prior knowledge of public pathway databases. Biocarta, Reactome, KEGG and Pathway Interaction Database information was retrieved using the R package rBiopaxParser. This we used to identify pathways of differentially abundant proteins in the proteomic dataset. We reconstructed a protein network containing a selection of the most relevant pathways, identified affected transcription factors via TRANSFAC database and performed a downstream determination of corresponding target genes. The downstream analysis we compared with the upstream analysis which determined upstream transcription factors and pathways of genes corresponding to differentially expressed transcripts identified in the RNA-Seq data set. Finally, we performed an overlap analysis with the objective of evaluating the method and prior knowledge incorporation. We used this approach to integrate time-dependent MS/MS (tandem mass spectrometry) and corresponding RNA-Seq data sets generated by stimulation of human B cell receptors. With our method an extensive parallel and time-dependent analysis of signalling events was feasible. Boston, MA June 17-19, 2014 171 Poster Abstracts #103 : Drug efficacy and side effects are strongly associated with its ability to reverse gene expression abnormalities in a mouse model of dyslipidemia Allon Wagner1 , Noa Cohen1 , Thomas Kelder2 , Elad Liebman3 , David M. Steinberg4 , Marijana Radonjic2 , Eytan Ruppin1,5 1 2 3 4 5 Blavatnik School of Computer Science, Tel-Aviv University, Israel Microbiology and Systems Biology, TNO, Zeist, The Netherlands Department of Computer Science, University of Texas at Austin, USA Department of Statistics and Operations Research, Tel Aviv University, Israel The Sackler School of Medicine, Tel Aviv University, Israel High-throughput gene expression data have proven invaluable in studying human disease, and are expected to become a principal element of future personalized medicine. At present, however, patients’ diagnosis and response to treatment are judged largely through measurements of disease-relevant physiological markers, mostly in blood and urine. The association between the molecular and the physiological manifestations of the disease has nevertheless not been investigated in a systematic, rigorous manner. Here, we study this fundamental relationship in a LDLR-/- mouse model of dyslipidemia that was subject to various treatments. We find that: (a) the deviations from the healthy state to the disease state on the transcriptomic (gene expression) and physiological levels are strongly correlated. (b) The restoration of individual physiological disease markers to their baseline values following most drug treatments is strongly correlated with a restoration of the transcriptome back to its normal baseline. (c) Physiological markers that are known to be associated primarily with a certain tissue are correlated with transcriptomic changes occurring in this tissue, and, importantly, (d) treatments that induce considerable ‘non-restorative’ transcriptomic alterations are associated with physiological side-effects. Overall, these results highlight the importance of searching for drugs that are able to restore the global cellular state back to its healthy norm, rather than rectify particular disease phenotypes. Such drugs are expected to be more effective and to have lesser side-effects. 172 International Conference on Systems Biology of Human Disease #104 : Network-based association of hypoxia-responsive genes with cardiovascular diseases Ruisheng Wang, William Oldham, Joseph Loscalzo Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School Oxygen is indispensable for cellular viability and function. Low cellular oxygen content (hypoxia) induces a number of molecular changes at different levels to activate regulatory pathways for increasing oxygen supply and optimizing metabolism under stress conditions. Hypoxia is involved in the pathobiology of many diseases, such as heart failure, stroke, and myocardial infarction. Although the complex association between hypoxia and cardiovascular diseases (CVD) has been recognized for some time, there are few studies investigating their biological link at a systems-level. In this study, we integrate hypoxia genes and CVD genes into the human interactome and explore the relationships between hypoxia and CVD using a systems biology approach. The analyses from macroscale to microscale all indicate the presence of interesting close relationships between hypoxia genes and CVD genes. Moreover, we find that hypoxia genes play significant bridging roles in connecting different CVDs. A few interesting signaling pathways and functional modules have been identified as common to both hypoxia stress response and CVDs. These signaling pathways and hypoxia-CVD functional modules provide new insights into the role of hypoxia in cardiovascular biology and disease. Boston, MA June 17-19, 2014 173 Poster Abstracts #105 : Classifying Cancer Types for Treatability Using PPI Network Structure Kathleen Wilkie1 , Michael La Croix2 , Philip Hahnfeldt1 1 2 Center of Cancer Systems Biology, GRI, Tufts University School of Medicine, Boston, MA, USA Department of Mathematics, Massachusetts Institute of Technology, Cambridge, MA, USA The existence of a correlation between cancer protein-protein interaction (PPI) network degree-entropy and cancer 5-year survival probability has been shown. We investigate the correlation between epidemiological survival data and the molecular details of specific cancers further through higher levels of degree-connectedness and through the evolution of cancer network connections compared to a random graph evolution. By comparing the cancer networks to random graphs of the same size, we create a new metric that correlates with 5-year survival and may act as a classifier. The correlative relationship suggests that our method is able to surpass individual variabilities such as cancer site, disease stage, and structural features of the PPI network, through the use of proxy measurements: 5-year survival data and our entropy-based metrics. Our findings suggest that the underlying structure of cancer protein-protein interaction networks may be used to classify cancer types into two groups: those that are more treatable for various reasons and those that are not. Furthermore, using spectral analysis techniques we present a new method to identify potential drug targets. Eigenvector centrality can be used to provide a ranked order to a subset of the vertices in a graph. To cover the whole network, we consider several of the largest magnitude eigenvalues and their corresponding eigenvectors found from the network’s associated adjacency matrix. Highest ranked vertices are considered potential drug targets and both individual and combination targets are presented. Drug targets are assumed to act by blocking the entire protein (vertex deletion) and the effects on residual network structure are discussed. 174 International Conference on Systems Biology of Human Disease #106 : Network algorithms dicover dysregulated pathways in RNAi screens Jennifer Wilson, Sara Gosline, Ernest Fraenkel, Doug Lauffenbruger Biological Engineering High-throughput, rna-interference (RNAi) screens are a powerful technique for studying the significance of gene mutations and deletions in cancer, yet high rates of false positives and false negatives limit the interpretation of these screens. More specifically, a gene-interference reagent’s performance may vary across replicates, and reagents targeting against the same gene may result in different phenotypic responses. To account for this uncertainty in a reagent’s performance, we apply a data integration approach to identify relevant pathways and better interpret screen hits. Our tool, the Simultaneous Analysis of Multiple Networks (SAMNet) uses RNAi screening and mRNA expression data with a custom interactome and mathematical optimization to find closely related sub-networks of screening targets. Further, we apply a randomization routine to look for targets that are specific to the experiment and not solely a result of study bias in the interactome. We’ve used SAMNet on an in vivo model of Acute Lymphoblastic Leukemia (ALL) to build sub-networks that identify differential and common pathways between the in vivo and in vitro experimental settings. Further, after running the optimization on randomized data, we enrich for targets that are specific to the ALL model. From these enriched targets, we can select novel genes for experimental validation and determine relevance of these genes to the in vivo development of ALL. Interpreting data in a network context identifies unique pathways not identified in either dataset alone. Further, this context provides higher confidence to identify true positives and we postulate that hidden targets from these pathways are candidate false negatives. Reducing false positives and false negatives improves data interpretation and the design of future experiments. In a therapeutic-development context, this analysis is useful for understanding mechanisms of disease, identifying biomarkers for diagnosis, understanding pathways for drug resistance, and hypothesizing potential combinatorial therapeutics. Boston, MA June 17-19, 2014 175 Poster Abstracts #107 : Cell-to-cell variability in NF-κB activation and chromatin at the HIV promoter contribute to noise in the reactivation of latent HIV Victor Wong1 , Kathryn Miller-Jensen2 1 2 Department of Molecular, Cellular and Developmental Biology, Yale University, New Haven, CT Department of Biomedical Engineering, Yale University, New Haven, CT Latent HIV infections are transcriptionally silent and therefore invisible to highly active antiretroviral therapies (HAART) and the host immune system. Consequently, latency is a barrier to the complete eradication of HIV. One promising therapeutic strategy is to purge the latent cellular reservoir by systematically activating latent HIV with latency-reversing agents (LRAs). However, recent observations demonstrate that induction of latent HIV provirus is stochastic in patient samples despite maximum T cell activation, suggesting that this strategy will be ineffective due to a failure to fully activate the latent reservoir. This approach may be further limited by noise in the kinetics of HIV activation, which may exceed the duration in which LRAs are effective. A better understanding of the dynamics and sources of noise in HIV induction is necessary to optimize this strategy. The activation of key host transcription factors, such as NF-κB, and the reversal of repressive chromatin at the HIV promoter are critical for the reactivation of latent HIV, but it remains unclear how these factors contribute to noise in HIV induction. We have used time-lapse microscopy to the study the kinetics of HIV induction at a single-cell level in two Jurkat T cell line HIV latency models in response to tumor necrosis factor (TNF), an activator of NF-κB signaling. We find that activation of HIV in both models exhibit substantial noise in the onset of activation and the kinetics of activation, and that the strength of this noise depends on the integration site. Co-stimulation with TNF and trichostatin A (TSA), a histone deacetylase (HDAC) inhibitor, decreases mean activation time and noise in activation in both models, suggesting that chromatin at the integration site contributes to transcriptional noise. Finally, by quantifying cell-to-cell variability in NF-κB activation using an immunostain, we find that noise in the NF-κB signaling pathway may contribute to noise in HIV activation. 176 International Conference on Systems Biology of Human Disease #108 : In Vivo Systems Analysis Identifies Cytokine Drivers of Neurodegeneration in Alzheimer’s Disease Levi Wood1,2 , Bradley Hyman2,3 , Douglas Lauffenburger4 , Kevin Haigis1,2 1 2 3 4 Department of Pathology, Massachusetts General Hospital, Boston, MA Harvard Medical School, Boston, MA Department of Neurology, Massachusetts General Hospital, Boston, MA Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA Alzheimer’s disease (AD) and related dementias are estimated to afflict more than 35 million people worldwide. The traditional approach to developing AD therapeutics has been to intervene in the processing and extracellular accumulation of amyloid beta (Aβ). Nevertheless, therapies based on the “amyloid hypothesis” have shown minimal efficacy in patients, suggesting that the activity of Aβ represents only one aspect of the complex pathogenesis of AD. Since the initial and ongoing physiologic response to Aβ involves microglial and astrocyte immune responses, we hypothesized that glial cytokines may play a role in promoting AD pathogenesis. Here, we show that a computational modeling analysis correlating cytokine protein expression to quantified pathologic disease state was able to identify a signature that reliably distinguished postmortem AD brain tissues from nondemented control brains. By comparing expression in different regions of the brain within individual patients, we were able to identify the cytokines (TNF-α, VEGF, IL-12, and IL-5) that were most up-regulated in the most degenerative brain region. By applying these cytokines to primary neuron cultures, we determined that TNF-α and IL-12, both of which have previously been connected to AD pathogenesis, promoted neuronal death. Surprisingly, we also found that VEGF promoted neuronal death, although only in the presence of Aβ. This synthetic death phenotype was commensurate with changes in phosphorylation of multiple signaling pathways that govern cell fate and could be abrogated by a small molecule inhibitor of VEGFR. These data implicate pro-death cytokine signaling in the pathogenesis of AD and, more broadly, demonstrate that multi-factorial computational methods can identify novel therapeutics leads for the disease. Boston, MA June 17-19, 2014 177 Poster Abstracts #109 : Exploring the cellular responses to short pulses of TNF Robin Lee1,2 , Mohammad Quasaimeh3 , Xianfang Xia1,2,4 , Kate Savery1 , David Juncker3 , Suzanne Gaudet1,2,4 1 Department of Cancer Biology and Center for Cancer Systems Biology, Dana-Farber Cancer Institute, Boston, MA, USA 2 Department of Genetics, Harvard Medical School, Boston, MA, USA 3 Department of Biomedical Engineering, McGill University, Montreal, Canada 4 BBS Program, Harvard Medical School, Boston, MA, USA How do cells make decisions? We are particularly interested in the cellular responses to the inflammatory cytokine Tumor Necrosis Factor (TNF). TNF has been implicated in a variety of diseases. More intriguingly, TNF can promote cell death and cell survival at the same time. It promotes survival by activating the transcription factor NF-kB while it promotes cell death by activating the caspase family of intracellular proteases. In most studies of TNF-induced signaling and cell death, cells are exposed to TNF continuously. However, in vivo, exposure to cytokines such as TNF can be of short duration. By using a simple microfluidic system and monitoring the real time response of cells expressing a fluorescently-tagged copy of the NF-kB subunit RelA, we established the concentration-dependent minimal duration of TNF treatment that induces NF-kB mediated transcription in HeLa cells. We found that while a single 30 second pulse is sufficient for full activation of NF-kB with 100 ng/ml TNF, a 1 minute pulse is needed for full activation with 10 ng/ml TNF. Strikingly, we observed that the duration of a TNF pulse also has a great impact on the timing and percentage of cell death. A single TNF pulse of 1 minute induces as much cell death as continuous treatment for 10 hours, but less death is seen with a pulse of 1 hour. 178 International Conference on Systems Biology of Human Disease #110 : A Multiscale-Modeling Approach Towards Understanding the Biological Response to Stress Nilgun Yilmaz1 , Hans V. Westerhoff2,3 , Andrzej Kierzek4 , Nick Plant4 1 VU Amsterdam, FALW, Molecular Cell Physiology, Amsterdam, The Netherlands. VU Amsterdam, FALW, Molecular Cell Physiology, Amsterdam, The Netherlands. 3 Manchester Centre for Integrative Systems Biology, Manchester Interdisciplinary Biocentre (MIB), Manchester, UK 4 Centre for Toxicology, Faculty of Health and Medical Sciences, University of Surrey, Guildford, Surrey, GU2 7XH, UK. 2 Biological systems are routinely exposed to external challenge, such as the stress response where each challenge causes a transient spike in blood cortisol. It is clear that prolonged or repeated stress is associated with the development of pathophysiology; for example, high stress police work is associated positively with the development of metabolic syndrome. It is both an important biological question, and challenging modeling scenario, to understand how these metabolic networks function efficiently, balancing both the pharmacodynamic (physiological effect of the challenge on the biological system) and pharmacokinetic (metabolic response to the challenge, leading to a return to homoeostasis) responses to stress. This topic has been covered in our previously published small-scale kinetic model showing an interaction of the stress hormone cortisol with its two cognate receptors, the high affinity glucocorticoid receptor (GR) and the low affinity pregnane X-receptor (PXR) [1]. In this publication, we were able to demonstrate that the interaction between these two receptors, plus the ligand promiscuity of PXR, were central to adapting to increasing frequency stress, potentially reducing the risk of pathologies associated with chronic stress. However, to fully understand the importance of this interaction network it is important to extend the model to represent whole body responses to stress signalling. In this work, we are working on development of a multi-scale model that can reproduce the complex biological responses to stress stimuli. To achieve this, a genome scale metabolic network (Recon2) was coupled to Petri net representations of the regulatory/gene signalling networks for both the interaction of cortisol with GR and PXR, and the circadian rhythmicity of cortisol expression. In this presentation, I would like to show you Recon2 is coupled with our small-kinetic network. If the results are available until the conference, I would like to share some results about the influence of circadian rhythmicity in stress response, as well. 1. Kolodkin et al. (2013) Nature Communications 4: 1792. Boston, MA June 17-19, 2014 179 Poster Abstracts #111 : Elucidating Mechanisms of Liver Metastasis: an All-Human Microphysiological Model to Investigate Disease Progression & Therapies Carissa L. Young1 , Sarah E. Wheeler2,3 , Amanda M. Clark2,3 , Donna B. Stolz2,3 , Alan Wells2,3 , Linda G. Griffith1 , Douglas A. Lauffenburger1 1 Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA Departments of Pathology, Cell Biology, Pharmaceutical Sciences, Bioengineering, University of Pittsburgh, PA 3 McGowan Institute for Regenerative Medicine, University of Pittsburgh, Pittsburgh, PA 2 Approximately 90% of cancer-associated mortality is a consequence of distant metastasis, whereby cells from the primary tumor invade or migrate from a primary site to secondary organs. These cancerous cells may either proliferate immediately or lay dormant (e.g. as pre-malignant micrometastases) for years prior to detection. Clinically, undetected metastases have serious implications for cancer patients; exemplified by ∼33% of women whom suffer a metastatic relapse within 5 years following removal of breast cancer. Distant metastases are generally more resistant to treatments than the primary tumor, hence facilitates the need to develop improved therapeutic approaches rationally designed based on the molecular pathophysiology within the metastatic microenvironment. To understand how cellular network deregulation underlies complex disease states, such as cancer, we implemented systems-based approaches to elucidate human-specific cellular crosstalk between tumor and hepatic tissues. The integration of primary cell sources, computational and systems biology approaches, as well as emerging biomaterials strategies and microfabrication processes promoted novel experiments and analyses to investigate breast cancer metastasis. We recapitulated the fundamental physiologic functions and conditions of the liver hepatic niche, including multi-cellular composition, metabolism, and protein production. Specifically, we determined how changes in the populations of multiple cell types, in the presence or absence of MDA-MB-231 cancer cells, regulate the response to inflammatory signals, i.e. cytokines, chemokines, and growth factors while corroborating the effects of cell function and viability across a broad spectrum of metrics. Statistical techniques, e.g. PCA and PLS – led to novel insights attributed to cytokine profiles and signaling networks, providing biological insights of plausible signatures for early metastatic disease. Collectively, the all-human microphysiological composition of our experimental system enables the development and validation of predictive models for cell/tissue behavior relevant in drug discovery. 180 International Conference on Systems Biology of Human Disease #112 : Detecting and Remembering Gut Dysfunction in Biomimetic Microfluidic Devices with Living Bacterial Diagnostics S. Jordan Kerns1,2 , Jonathan W. Kotula1,2 , Katia Karalis2 , Jeffrey C. Way2 , Donald Ingber2 , Pamela A. Silver1,2 1 2 Department of Systems Biology, Harvard Medical School Wyss Institute for Biologically Inspired Engineering The mammalian gut is a dynamic community of symbiotic microbes that closely interact with the host and have a profound impact on human health such as inflammatory/autoimmune disease, obesity, and diabetes. However, our ability to non-destructively interrogate the gut is limited. We have engineered E.coli and S.typhimurium that sense and record environmental stimuli, and have demonstrated that engineered bacteria with such memory systems can survive and function in the mammalian gut. The challenge now is to harness the powerful endogenous sensing and signaling networks of enterobacteria like E.coli and S.typhimurium to record specific disease biomarkers. However, building biological memory systems capable of acting as clinically relevant diagnostic agents requires iterative engineering that is impractical with animals and incompatible with cell-based models that fail to recapitulate key aspects of gut physiology. Biosimilar ‘Gut-on-a-Chip’ microfluidic devices mimic the mechanical and physiological microenvironments of the mammalian gut and can be applied as an ex vivo platform to accelerate the development of nondestructive, living bacterial diagnostics of gut enteropathy. Specifically, synthetic bacterial memory systems activated by upstream endogenous bacterial gene-promoters are shown here to detect and respond to disease-like states induced in Gut-on-a-Chip devices. Boston, MA June 17-19, 2014 181 Poster Abstracts 182 International Conference on Systems Biology of Human Disease General Information General Information Joseph B. Martin Conference Center Ground Floor women men elevators courtyard stairs stairs lobby stairs Amphitheater First Floor elevators stairs open to courtyard Amphitheater Balcony open to below stairs stairs 184 International Conference on Systems Biology of Human Disease HMS Campus Map Boston, MA June 17-19, 2014 185 General Information Area Street Map 186 International Conference on Systems Biology of Human Disease