ABSTRACT. Quantitative Systems Pharmacology (QSP) has been described as “the quantitative analysis of the dynamic interactions between drug(s) and a biological system to understand the behaviour of the system as a whole, as opposed to the behaviour of its individual constituents,” (van der Graaf and Benson, Pharm Res, 2011). In 2011, an NIH working group on QSP issued a white paper espousing the idea that improving the success of pharmaceutical research and development can in part be accomplished by reinvigorating the field of pharmacology “by introducing concepts, methods and investigators from computational biology, systems biology and biological engineering, thereby allowing modern pharmacologists to apply systems-level ideas to practical problems in drug development. QSP has deep roots in classical pharmacology and physiology but adds a molecule and systems-level approach that allows drug responses to be studied in the context of increasing knowledge of the complex and subtle interconnectedness of signaling, transcriptional and metabolic networks, as well as the variation in individual patients arising from differences in genetics and environment.” QSP has been gaining traction in the biopharmaceutical industry, in academic departments and professional organizations in pharmaceutical sciences, and recently, with regulatory agencies as well (Peterson and Riggs, CPT-PSP 2015). However, further advancing this field requires continued close interaction of the QSP community with the systems biology and biological engineering communities. As such, we propose that a workshop on QSP at ICSB, highlighting how systems biology approaches can be leveraged in the context of systems pharmacology, can enrich both fields and further collaborative work in this area that is proving increasingly valuable in pharmaceutical research and development.

Schedule: Quantitative Systems Pharmacology - Mon Aug 7, 8:30am-noon

8:30 Opening remarks by Valeriu Damian (GlaxoSmithKline)

8:45 Quantitative Systems Pharmacology and Toxicology - QSP & QST : An overview by Valeriu Damian (GlaxoSmithKline)

9:15 Bridging the gap between Systems Biology and QSP: Application on modeling the Complement pathway and evaluating treatments for autoimmune diseases by Loveleena Bansal (GlaxoSmithKline) 

9:45 Break

10:00 Complementary Systems Pharmacology approaches to dissecting PI3K-inhibitor dependent GI toxicity (part 1) by Jonathan Wagg (Roche) 

10:30 Complementary Systems Pharmacology approaches to dissecting PI3K-inhibitor dependent GI toxicity (part 2) by Kapil Gadkar(Genentech) 

11:00 Integrated transcriptomics and mathematical modeling analysis for quantitative predictions of drug-induced toxicity. by Eric Sobie (Mount Sinai School of Medicine)

11:30 Panel Discussion: Chellenges in bridging Systems Biology with Quantitative Systems Pharmacology and Toxicology by All speakers

12:00 Session ends

Participants List: 

Chair Valeriu Damian, GlaxoSmithKline damiav01@gsk.com

Co-Chair Eric Sobie, Mount Sinai School of Medicine, eric.sobie@mssm.edu

Authors: 

Loveleena Bansal, GlaxoSmithKline, loveleena.x.bansal@gsk.com

Jonathan Wagg, Roche, jonathan.wagg@roche.com

Kapil Gadkar, Genentech, gadkar.kapil@gene.com

ABSTRACT. Computational analysis of molecular interaction networks has become pervasive in systems biology. Nearly every publication that uses network analysis includes a visualization of a graph in which the nodes and edges are laid out in two dimensions. Several systems implement methods for creating such layouts. Despite these advances, interdisciplinary research teams in network biology face several challenges in sharing, exploring, and interpreting computed networks in their collaborations.

GraphSpace is a web-based system that provides a rich set of user-friendly features designed to enhance network-based collaboration:

• Users can upload richly-annotated networks, irrespective of the algorithms or software used to generate them.
• Users can create private groups, invite other users to join groups, and share networks with group members.
• A user may search for networks that contain a specific node or edge, or a collection of nodes and edges.
• A powerful layout editor allows users to efficiently modify node positions, edit node and edge styles, save new layouts, and share them with other users.
• Researchers may make networks public and provide a persistent URL in a publication, enabling other researchers to explore these networks. 

This tutorial will provide an in-depth introduction to GraphSpace. Attendees will receive hands-on training on the GraphSpace web interface and how to incorporate programmatic interaction with GraphSpace into their network analysis projects.

ABSTRACT. COPASI is a software application for simulation and analysis of biochemical networks. It is developed jointly by the groups of Pedro Mendes, Ursula Kummer, Sven Sahle, and Stefan Hoops and is freely available for academic and commercial use.

COPASI's current features include stochastic and deterministic time course simulation, steady-state analysis (including stability), metabolic control analysis, elementary mode analysis, mass conservation analysis, import and export of SBML level 1 - 3, optimization, parameter scanning and parameter fitting. It runs on MS Windows, Linux, and Mac OS X

We will use COPASI to explain how the modelling, simulation and computational analysis of biochemical systems works. We will also critically evaluate the limitations of different simulation methods.

ABSTRACT. As recent pandemics, such as the Zika and Ebola outbreaks have shown, diseases spread very fast in today’s interconnected world, making public health an important research area. Some of the basic questions are: How can an outbreak be contained before it becomes an epidemic, and what disease surveillance strategies should be implemented? These are challenging problems at the interface of dynamical systems, graph theory, data mining, machine learning, high performance computing, theoretical computer science, economics and statistics. In this tutorial, we provide an overview of the state of the art in mathematical and computational epidemiology, which have typically not been studied from a multi-disciplinary perspective. The tutorial is suitable for both novice and expert researchers, and will be at a level that is accessible to most ICSB attendees who work in these areas.

ABSTRACT. Modern systems biology often requires an integrated computational and experimental approach. Countless tools and methods have been developed to facilitate research, but many lack engaging and approachable tutorials. This makes it more difficult for other scientists and students to learn how to use those tools in an effective way. Working through a specific comprehensive task that explicitly walks through acquiring relevant data, performing analysis, and visualizing results has been shown to be a more effective technique. In this workshop we will work through a series of modules we have developed to teach cancer systems biology, which have received positive student feedback, as a springboard and general template for creating tutorials of your own research tools and methods.

ABSTRACT. Beyond the use of small molecule inhibitors as tool compounds to study a variety of biological processes, assays of cellular response to drugs are a fundamental aspect of the development and characterization of therapeutic molecules and the investigation of drug mechanism of action. However, drug response measurements and their analysis are not as trivial as one thinks and large-scale efforts across panels of cell lines have been plagued by inconsistencies. In addition, the quantification of drug combinations has been a topic of controversy for decades, with multiple methodologies leading to conflicting conclusions. These issues have motivated recent efforts to advance the methodology and theory for drug-response assays. In this workshop, we will present improved experimental and computational methods to generate reproducible dose-response measurements across cell lines, as well as theoretical approaches to quantify the sensitivity of cells to single drugs and drug combinations.

ABSTRACT. We present StochSS: Stochastic Simulation as-a-Service, an integrated development environment for modeling and simulation of deterministic and discrete stochastic biochemical systems. An easy to use WebUI enables researchers to quickly develop and simulate biological models on a desktop or laptop, which can then be expanded or combined to incorporate increasing levels of complexity. As the demand for computational power increases, StochSS is able to seamlessly scale by deploying cloud computing resources. The cloud computing facilities also make it possible to deploy StochSS as a multi-user software as-a-service (SaaS) environment with the capability to share and exchange models via a public model repository. StochSS currently supports simulation of ordinary differential equations and well-mixed discrete stochastic models, as well as parameter estimation of discrete stochastic models and efficient mesoscale simulation of spatial stochastic models. StochSS is available for download at www.StochSS.org.

ABSTRACT. Microbes are by far the most dominant forms of life on earth. In every imaginable habitat, they form complex communities that carry out diverse functions. Microbial communities drive the geochemical cycling of diverse chemicals and through these activities shape the earth’s climate and environment. They are also intimately tied to human physiology and health. Members of each microbial community may compete for resources, collaborate to process the resources or to cope with stress. They communicate with each other by producing and responding to signaling molecules. And they innovate by exchanging genetic materials. These interactions raise fundamental questions regarding the evolutionary and ecological forces that shape microbial consortia.  Our lab has adopted a combination of quantitative biology and synthetic biology to explore these questions. We engineer gene circuits to program dynamics of one or more Escherichia coli bacterial populations and use them to examine questions in cellular signal processing, evolution, ecology, and development. Analysis of these systems has provided insights into bacterial tolerance to antibiotics, developmental pattern formation and scaling, as well as strategies to use bacteria to fabricate functional materials by exploiting programmed self-organization.

ABSTRACT. The adoptive transfer of T cells expressing chimeric antigen receptors (CARs) has demonstrated clinical efficacy in the treatment of advanced cancers, with anti-CD19 CAR-T cells achieving up to 90% complete remission among patients with relapsed B-cell malignancies. However, challenges such as antigen escape and off-tumor toxicity limit the long-term efficacy and safety of adoptive T-cell therapy. Here, I will discuss the development of next-generation T cells that can perform Boolean logic signal processing to increase the robustness and specificity of therapeutic T cells. This presentation will highlight the potential of synthetic biology in generating novel mammalian cell systems with multifunctional outputs for therapeutic applications.

ABSTRACT. Our lab developed mass spectrometry-based methods to enable high-throughput metabolomics – enabling detection of 300-800 compounds in up to 2000 samples per day. These methods enabled, for example, mapping of the so far uncharted gene-metabolite associating network¹ or genome-wide discovery of novel enzyme activities². Since cellular metabolism responds directly and indirectly to environmental changes and regulatory events, changing metabolite concentrations are typically difficult to interpret. To delineate the different regulatory events in dynamic experiments, we recently developed near real-time metabolomics³. By following metabolome responses for minutes up to several hours at a resolution of a few seconds, this method allows to separate immediate kinetic and allosteric regulation from longer term processes. Here I will focus on so far unpublished results on the identification of the regulation processes that determine the decision to grow in E. coli. Specifically, we follow the metabolome responses during intermittent carbon feeding from non-growth supporting frequencies to growth. Surprisingly, even minute amounts of glucose gush through central metabolism all the way to building blocks of macromolecules. The decisive element for cell division is the balance between synthesis and degradation of a single protein.

¹ Fuhrer T, Zampieri M, Sevin DC, Sauer U & N. Zamboni. 2017. Genome-wide landscape of gene-metabolome associations in E. coli. Molecular Systems Biology 13: 907.

³ Sevin DC, Fuhrer T, Zamboni N & U Sauer U. 2016. Nontargeted in vitro metabolomics for proteome-scale identification of novel enzymes in E. coli. Nature Methods 14:187-194.

³ Link H, Fuhrer T, Gerosa L, Zamboni N & U. Sauer. 2015. Real-time metabolome profiling reveals dynamics and regulation of the metabolic switch between starvation and growth. Nature Methods 12: 1091-1079.
 

ABSTRACT. Large-scale genomic studies are providing unprecedented insights into the molecular basis of cancer, but it remains challenging to leverage  this information for the development and application of therapies. We have performed an integrated analysis of the molecular profiles of large number of  primary tumours and  cancer cell lines, along  with the response of the cell lines to anti-cancer compounds. Integration of  this data with various sources of prior knowledge, in particular signaling pathways and transcription factors, points at molecular processes involved in resistance mechanisms. Our own analysis as well as the results of  a crowdsourcing effort (DREAM challenge) reveals that  prediction of drug efficacy is far from accurate, implying important limitations for personalised medicine. I will argue than an important aspect that needs to be further studied is the dynamics of signaling networks and how they response to drug treatment. I will show how applying logic models, trained with phosphoproteomic measurements upon perturbations, can further improve our understanding of the molecular basis of drug resistance, thereby providing new treatment opportunities not noticeable by an static molecular characterisation.

ABSTRACT. Establishing the core requirements of cellular life is an important challenge of biology. The question of the minimal set of biochemical functions necessary for a cell to grow and replicate has been studied from a number of angles for 20 years, culminating in the recent construction of the first "minimal cell" by synthetic biologists [1]: Starting from the pathogenic bacterium Mycoplasma mycoides, several cycles of genome design and assembly led to the removal of all genes not required for robust growth under optimal conditions. The resulting organism, Syn3.0, contains only 473 genes in a 531 kbp genome - less than any other known autonomously replicating cell. This minimal genome opens up the opportunity to, for the first time, understand all gene functions within a living cell - and cast them into a complete computational model encompassing all cellular functions.

As a first step towards this goal of a complete computational model of the minimal cell, we present here the metabolic reconstruction of Syn3.0, cast into a flux-balance analysis (FBA) model. Combining the genomic information of Syn3.0 with available experimental information on Mycoplasma mycoides, we have assembled a metabolic model encompassing around 150 genes. This model allows us to probe and analyze the metabolic properties of the minimal cell. In particular, it allows to rationalize experimental transposon insertion data [1] which suggests that disrupting some genes affects the cell not as strongly as disruption of other genes. Looking forward, the metabolic reconstruction will serve as a foundation for more comprehensive in silico models of Syn3.0, integrating all of its transcription, translation and metabolic processes.

[1] Hutchison C. A. et al. Science, 2016, 351, aad6253.

ABSTRACT. Sub-cellular compartmentalization is used by all eukaryotes and some prokaryotes as a means to create favorable microenvironments for various metabolic reactions. These compartments can concentrate enzymes, separate competing metabolic reactions, and isolate toxic intermediates in metabolic pathways. Such advantages have been recently harnessed by metabolic engineers to improve the production of various high-value chemicals via compartmentalized metabolic engineering. However, one challenge in compartmentalized metabolic engineering is to determine key metabolite level in these compartments. Conventional techniques such as metabolomics analysis and transcription-based biosensors can only reflect cytosolic metabolite concentration instead of compartmental metabolite concentration. To this end, we developed a synthetic biosensor to determine a key metabolite, i.e., acetyl-CoA, in a representative compartment of yeast, i.e., peroxisome. This synthetic biosensor used highly efficient enzyme re-localization via PTS1 signal peptides to construct a metabolic pathway in the peroxisome, which solely converted peroxisomal acetyl-CoA to polyhydroxybutyrate (PHB) via three enzymes, phaA phaB and phaC. By quantifying the PHB level in yeast, we successfully determined peroxisomal acetyl-CoA level under various culture conditions. We next performed a proof of concept for our biosensor by screening a library of single knockout yeast mutants and identified one yeast mutant, ΔRPD3, which had elevated level of peroxisomal acetyl-CoA compared to wild type yeast. We expect our synthetic biosensors can be widely used to deepen our understanding of sub-cellular compartmental metabolism and facilitate the “design-build-test” cycle of compartmentalized metabolic engineering.

ABSTRACT. In its efforts to produce economically feasible bioethanol from non-edible plant parts, the biofuel industry is interested in metabolically engineered biomass with reduced recalcitrance. Recalcitrance, the resistance of biomass to enzymatic degradation, limits the accessibility of enzymes to plant sugars and lowers the efficiency of ethanol fermentation. Recalcitrance is due to the heteropolymer lignin, which is interwoven with cellulose and hemicellulose in plant cell walls. Therefore, reducing lignin content and altering its composition in target plants is one of the primary goals of biofuel research. Interestingly, industries associated with textile production and the synthesis of organic compounds have identified lignin as a valuable starting compound for a novel processes and are interested in maximizing it. The branched structure of the lignin biosynthetic pathway exemplifies a nonlinear system whose functionality is difficult to understand without a computational modeling approach. We thus constructed a dynamic model of the pathway in the model grass Brachypodium distachyon, using 13C9-labeling data from control and perturbation experiments for calibration. Unlike dicots, where phenylalanine is the sole precursor of monolignols, monocots like Brachypodium may use either phenylalanine or tyrosine, with the source affecting the ultimate monolignol composition. Although lignin synthesis has been studied for some time, the dynamics of the pathway is not entirely known, and the different monolignol compositions are puzzling. Preliminary analysis demonstrated that a single-compartment model could not explain the differences either. We thus formulated a dynamical model containing enzymes of the phenylalanine pathway at external surface of the ER and those of the tyrosine pathway in the cytosol. Intermediate metabolites are preferentially channeled within compartments but can be shuttled from one site to the other. With this design, the model results became consistent with experimental observations and can explain different monolignol compositions resulting from the preferential incorporation of either phenylalanine or tyrosine.

ABSTRACT. With ultra-precise genome editing tools at our disposal, the life sciences will shift from one-factor-at-a-time type of experiments to an ever increasing need to design complex non-intuitive manipulations involving simultaneous changes at multiple loci. In principle, integration of omics data and systems biology models would provide the means for optimizing knowledge gain through rational target selection and experimental design. They are not leveraged effectively, however, due to a lack of readily available tools to rapidly access and analyze public and private data to design genetic and experimental manipulations. With this project we aim to make a broad spectrum of omics data useful to biotechnology and life science research by integrating systems biology with design in a one-stop platform that will serve a variety of application areas, ranging from industrial biotechnology to agriculture and human health. A group of five academic partners (DTU, Chalmers, EMBL, EPFL and UMinho) will drive basic research on integrative, model-based omics data analysis to enable: (1) metagenomics-enabled design of novel enzymes and biochemical pathways, (2) omics data-driven design of cell factories for the production of chemicals and proteins, and (3) analysis and design of microbial communities relevant to human health, industrial biotechnology and agriculture. All research efforts will be integrated in an interactive, web-based platform available to both industrial and academic research. The platform will be composed of standardized and interoperable components and therefore easily extensible. This will accelerate the process of bringing cutting edge systems biology research into practice. A first iteration of the platform, which features an intuitive user interface to data-constrained, genome-scale metabolic modeling that is based on interactive pathway visualizations, is already available at http://dd-decaf.eu.

ABSTRACT. People dying of infections that are not treatable anymore due resistance of pathogenic bacteria to various of antibiotics, have caused the urgent need to develop production of novel types of antibiotics. Streptomyces isolates are receiving considerable attention, especially due to unique antibiotic activity of their secondary metabolites. The genus Streptomyces is recognized to produce 40% of all known compounds with antibiotic activity, and it has been successfully used for heterologous expression.

The aim is to develop a mathematical and computational framework for the improvement of heterologous expression of gene clusters encoding for secondary metabolite pathways in Streptomyces coelicolor (S.coelicolor) strains. We have built a genome-scale metabolic model of S.coelicolor through extending the reaction universe of the existing iMK1208. The reaction universe has been extended by adding information from the previous iMA789 model. Additionally, the KEGG pathways have been used to add missing reactions and metabolites.

The resulting genome-scale metabolic model is then iteratively corrected and refined, according to internal consistency criteria and by comparing its predictions to experimental data, such as growth, gene-knockout, uptake and secretion rates. The resulting model is mapped to the KEGG namespace. This increase the value of the model, by simplifying further modifications and integration with the extensive pathway and reaction universe within the KEGG database.

ABSTRACT. Systems biology aims to understand how genotype influences phenotype. This requires comprehensive mechanistic models such as whole-cell models. Genomics has produced numerous datasets and predictors that could enable such models. However, these resources are often overlooked because they are distributed across many databases and manuscripts. To accelerate whole-cell modeling, we are developing KineticDatanator, a tool for aggregating data for models from databases and predictors. KineticDatanator (a) retrieves data; (b) identifies relevant data for models from similar organisms, molecules, reactions, and conditions; (c) merges data to estimate model parameters; (d) reviews data; (e) organizes data for model construction; and (f) tracks provenance. KineticDatanator contains two layers: (a) a core for retrieving, filtering, merging, reviewing, and recording data and (b) modules which use this core to aggregate data from specific resources. KineticDatanator’s core uses several tools to identify relevant data by chemical similarity (Open Babel), reaction similarity (E-zyme), species similarity (NCBI Taxonomy), and gene similarity (KEGG ORTHOLOGY). Currently, KineticDatanator includes modules for aggregating metabolite concentrations (ECMDB, YMDB), RNA concentrations (GEO), protein concentrations (literature), and kinetic parameters (SABIO-RK). Furthermore, KineticDatanator's modular architecture enables developers to contribute additional modules. We have used the SABIO-RK module to identify catalytic rate and affinity constants for 381 of 466 (82%) of the reconstructed metabolic reactions of Mycoplasma pneumoniae. 58% of the data was aggregated from Terrabacteria. 28% percent of the data was aggregated from observations of the identical reaction and 72% was aggregated from chemically-similar reactions. We plan to develop additional modules for additional data types such as promoter sites, protein localizations, DNA binding motifs, and RNA half-lives. Furthermore, we plan to develop a web-based interface to enable researchers to collaboratively aggregate data. We anticipate that KineticDatanator will accelerate whole-cell modeling and that these models will transform biological research.

ABSTRACT. Biology emerges from interactions between molecules, which are challenging to elucidate with current techniques. An orthogonal approach is to probe for "response signatures" that identify specific circuit motifs, which describe interactions globally. For example, bistability, hysteresis, or irreversibility are used to detect positive feedback loops. For adapting systems, which are ubiquitous in biology, such signatures are not known. Two different circuit motifs generate adaptation: negative feedback loops (NFLs) and incoherent feedforward loops (IFFLs), which specify different interactions and exhibit different biology. Based on exhaustive computational testing and mathematical proofs, we propose the first differential signatures: In response to oscillatory stimulation, NFLs but not IFFLs generically show i) 'refractory period stabilization' (robustness to changes in stimulus duration) or ii) 'period skipping'. Applying this approach to wild-type and mutant yeast, including a synthetic IFFL circuit, we identified the circuit dominating cell cycle timing. In C. elegans AWA olfactory sensory neurons, which are crucial for chemotaxis, we uncovered a Calcium-NFL leading to adaptation, difficult to find by other means, especially in wild-type, intact animals. These new response signatures allow direct access to the outlines of the wiring diagrams of adapting systems.

ABSTRACT. The BMP (bone morphogenetic protein) signaling pathway, critical in controlling diverse aspects of mammalian development, includes a dozen ligands and seven receptors that regulate a single biochemical output, the phosphorylation of Smad1/5/8 second messengers. Despite this seemingly redundant role for the BMP ligands, these ligands are not equivalent in all developmental contexts. Moreover, in the development of tissues including bone, heart, brain, and kidney, multiple BMP ligands are co-expressed. Indeed, it has been shown that ligand competition for shared receptors results in combinatorial signal processing. Specifically, cells do not respond to only the sum of two equivalent ligands, termed an “additive” response, but can be sensitive to the ratio of two inequivalent ligands, responding maximally to a balance, imbalance, or specific ratio of those ligands.

This framework provokes the general question of how the many BMP ligands combine with one another to control the activity of the BMP pathway. To address this question, we characterized the combinatorial properties of 14 BMP ligands by measuring responses to all possible 91 pairs. Using these data, we identify the distribution of integration modes, highlighting which BMP pairs combine additively and which pairs show a different, non-redundant behavior. Clustering the ligands based on their pairwise profile shows BMPs can be described in terms of equivalent groups. Within each cluster, ligands combine additively, while members of distinct clusters are integrated using a specific pairwise mode as previously described. Furthermore, mathematical models of BMP signaling indicate that receptor abundance can alter signal integration by modulating competition for ligand binding sites. Starting from this theoretical prediction, we show experimentally how changes in receptor profile alter the functionally equivalent clusters of BMP signals. These results should provide a more predictive understanding of how cells communicate using combinations of ligands and changes in receptor profile.

ABSTRACT. T cells provide cell mediated immunity in vertebrates against pathogens and diseases. During an immune response, naive CD4+ T cells activate and differentiate into effector or regulatory cell sub-types. Recent studies have shown that T-cell phenotypes exhibit plasticity that involves production of intermediate states, e.g., phenotypes co-expressing lineage specific transcription factors, and interconversion between previously thought terminally differentiated phenotypes, e.g., Th17 to Th1. The plasticity of T cell differentiation is regulated by cytokine microenvironment, intracellular signaling, and gene regulation, but the exact mechanism is unclear. Dynamic interplay of extracellular cytokines play an important role in regulation of CD4+ T cell differentiation plasticity. Recently, computational modeling emerged as an important tool to study the dynamics of biological systems, therefore can be used to gain insight into regulation of T-cell differentiation plasticity. We have developed a comprehensive logic-based computational model depicting regulatory mechanisms involved in CD4+ T cell differentiation. We obtained information of regulation of CD4+ T cell differentiation from literature and constructed the model in Cell Collective. The model consists of four master transcription factor regulators i.e. Tbet, GATA3, RORgt, and Foxp3, six STATs and other signaling molecules. The model incorporates 96 regulatory interactions among 38 components that include nine cytokines in extracellular environments. The model was simulated under 510 combinations of nine extracellular cytokines (environmental conditions). Each environmental condition was characterized by varying doses of active cytokines. Simulation and analysis results suggest ten potentially stable phenotypes, including four main types of differentiated CD4+ T cells, as well as hybrid phenotypes co-expressing lineage specific master regulators. In addition to already known phenotypes, we found novel phenotypes co-expressing lineage specific transcription factors. In conclusion, we predicted the optimal activity levels of extracellular cytokines that induce the production of novel CD4+ T cell phenotypes co-expressing lineage specific transcription factors.

ABSTRACT. High risk forms, human papillomaviruses (HPV) early promoter regulation is of paramount importance to understand the early phase of the infection and cancer evolution. The aim of this work is to develop a novel stochastic mathematical modeling framework, able to capture known biological mechanisms related to HPV early promoter regulation. The model includes modules designed to account for the transcriptional, post-transcriptional and translational regulation of E1, E2 early genes and E6 and E7 oncogenes, properly coupled to form the entire network. To appropriately model the post-transcriptional regulation, the major splicing sites and the splicing factor SRSF1 regulation were considered The Master Equation governing the model stochastic evolution was solved by means of the Gillespie algorithm and the stochastic behaviour was compared with the deterministic mean system behavior provided by a quasi-equilibrium approximation of the Master Equation. The model resulted to be able to fit a dataset with early promoter activities and to reproduce patterns qualitatively/quantitatively consistent with the known biology of the virus. Results also suggested that stochasticity plays a pivotal role in determining the dynamics of HPV gene expression. In particular, the combination of positive and negative feedback regulation of the early promoter generates stochastic bursts of gene expression, having amplitude and frequency modulated by, respectively, the relative strengths of the feedback loops and the post-transcriptional regulation by splicing. This latter mechanism is also responsible for a stochastic switch behavior that moves the system from normal infection to a condition in which E6 and E7 are overexpressed while E1 and E2 are no longer transcribed, consistent with HPV integration. The developed model appears as an important tool in predicting the early promoter regulation, useful to shed light in important and still elusive mechanisms and showing the pivotal importance of the intrinsic stochasticity in the HPV gene expression.

ABSTRACT. The proliferation of B lymphocytes to enable production of antibodies specific to invading pathogens is critical for an effective immune response. The prevailing hypothesis that B-cell fate decision-making is highly stochastic was based on timepoint measurements revealing high generational heterogeneity and probabilistic mathematical models that recapitulated population dynamics. Here, using multi-scale mechanistic computational modeling of the molecular network that is responsible for cellular decision making, and live cell microscopy of B-cell population expansion, we are able to quantify the contributions of both founder cell heterogeneity (extrinsic noise) and molecular stochasticity (intrinsic noise). Indeed, we identify non-genetic heterogeneity in founder cells as the key determinant of B-cell population dynamics, rather than probabilistic decision making. This means that contrary to previous models only a minority of genetically identical founder cells contribute the majority to the population response. We computationally identify, and experimentally confirm, non-genetic determinants of proliferative outcome. Our finding of the largely deterministic nature of B-lymphocyte responses renders the control of humoral immune responses more amenable to diagnostic and therapeutic development than previously thought.

ABSTRACT. Accurate chromosome segregation during mitosis requires that the sister kinetochores on each chromosome are stably attached to spindle microtubules prior to cell division. If one or more kinetochores are unattached, they delay cell division by activating the mitotic checkpoint, also known as the Spindle Assembly Checkpoint (SAC). The SAC is a complex signaling cascade uses many kinetochore proteins, signaling proteins, and their extensive phosphoregulation, to produce a diffusible ‘wait-anaphase’ signal. The objectives of the SAC cascade are two-fold: to maximize accurate chromosome segregation and to minimize mitotic delays. For the first objective, it is essential that a single unattached kinetochore generates a strong signal to delay anaphase onset. To meet the second objective, the large number of unattached cells present at the beginning of mitosis should not produce a proportionately large signal. How a nanoscopic kinetochore can generate a strong ‘wait-anaphase’ signal to inhibit anaphase over a 106-fold larger cellular volume, or why larger numbers of unattached kinetochores don’t produce a proportionately larger signal remain two fundamental questions in cell biology. The primary obstacle to answering these questions has been the localization of most of the signaling reactions within the nanoscopic kinetochore. Therefore, we engineered the ‘eSAC’ – a diffusible, kinetochore-independent, and quantifiable SAC activator. The eSAC ectopically activates the SAC by conditionally dimerizing the scaffold protein for SAC signaling with the kinase that activates the cascade. Using the eSAC, we conducted quantitative dose-response analyses of the SAC cascade in live cells. These analyses reveal two novel properties of the SAC cascade. We find that the recruitment of multiple SAC proteins by the signaling scaffold stimulates synergistic signaling. This enables a small number of scaffold molecules produce a disproportionately strong anaphase-inhibitory signal. However, many scaffold molecules signal concurrently, they compete for a limited cellular pool of SAC proteins. This frustrates synergistic signaling and modulates signal output. We propose that these two mechanisms institute automatic gain control – inverse, non-linear scaling between the signal output per kinetochore and the unattached kinetochore number, and thus achieve the two objectives of SAC signaling.

ABSTRACT. Traditional herbal medicine (THM) consists of multiple herbal ingredients, and the combination of these herbs is regarded as the key to their pharmacological efficacy. For example, maoto, a traditional Japanese medicine (Kampo) that is prescribed for influenza-like symptoms, contains four different herbs: Armeniacae Semen (AS), Glycyrrhizae Radix (GR), Cinnamomi Cortex (CC), and Ephedrae Herba (EH). While the effect of the major active ingredient “ephedrine” (from EH) has been well studied, the pharmacological action of maoto as a whole remedy remains unknown. In our previous study, seven major ingredients were derived from the constituent herbs and several hundred ingredients/metabolites were detected in rat plasma after maoto administration. This indicates that the pharmacological properties of maoto result from a complicated combination of multiple ingredients that may not be attributable to EH alone. To elucidate the combinatorial effect of maoto further, we compared pharmacological, metabolomic and transcriptomics properties among ephedrine, a mixture of major ingredients (toy-maoto) and maoto in a rat model of polyI:C-induced inflammation. Maoto ameliorated both disease symptoms and the surge in proinflammatory cytokines. Although ephedrine contributed largely to suppressing the acute cytokine surge, the inhibition of body weight loss caused by the mixture of herbs was not reproduced by ephedrine alone. In the plasma metabolome, maoto broadly affected lipid mediator responses and modulated the balance of proinflammatory and anti-inflammatory lipid mediators. By contrast, ephedrine played a major role in the early effects of maoto, including alteration of amino acids and metabolites related to the TCA cycle. In the lung transcriptome, ephedrine inhibited major proinflammatory factors, while toy-maoto and maoto affected broader pathways. In summary, we have demonstrated that the specific pharmacological properties of THM are exerted by a mixture of herbs but not by a single ingredient via the horizontal integration of pharmacological and molecular profiling.

ABSTRACT. With the advent of high-throughput technologies capable of delivering massive amounts of data, the demand for comprehensive panels of molecular markers of disease has increased correspondingly. Comprehensive panels that stem from whole-genome expression, transcriptome or other omics profiles (proteome, secretome, exosome, glycome, metabolome) are sought for every level of diagnostics, from preliminary screening for the presence or risk of a disease, to staging, response to treatment, progression or relapse. In this work, we explored the outcome of mass spectrometry (MS)-based proteomic experiments aimed at profiling the G1 and S cell cycle stages of cancer and nontumorigenic cells to identify functionally-related biomarker proteins that could be recommended for further exploration in clinical context.

The MCF7/ER+, SKBR3/HER2+ breast cancer and MCF10 non-tumorigenic cells were used as an experimental model system. Protein extracts from each cell line and cell cycle stage were analyzed by nano-LC-MS/MS, and the raw data were processed with the Discoverer and Mascot search engines. Each cell state enabled the identification of a substantial number of markers, totaling ~350 proteins with biomarker potential. Cluster analysis revealed that the top enriched biological processes that were represented by these proteins included three major categories related to DNA damage repair, oxidative stress and signaling, as well as several smaller categories pertaining to epithelial to mesenchymal transition, adhesion, response to various types of radiation, and regulation of cell proliferation and apoptosis. A number of proto-oncogenes and mutated proteins were identified, as well. To understand the power and challenges associated with the use of proteomic data in biomarker research, we will detail in our presentation the experimental design, the mass spectrometric and statistical protocols used for protein identification and validation, and the bioinformatics work-flow that was developed for data interpretation. The role of protein-protein interactions and gene regulatory networks in the development of biomarker signatures characteristic of a disease will be discussed, and the impact of driver mutations on aberrant cell cycle progression and cell proliferation will be addressed.

ABSTRACT. B. licheniformis has been successfully identified and deployed as an active producer of a number of industrially relevant products including surfactants and antibiotics. Furthermore, B. licheniformis’ ability to form spores makes it a potential candidate to be used as a biological control agent. Here, we report the complete genome sequences of two B. licheniformis strains Bac48 and Bac84 that were isolated from mangrove mud and microbial mat from Rabigh lagoon by the Red Sea. Using the 16S rRNA gene, Bac48 and Bac84 are 99% and 98% similar to B. licheniformis DSM 13, respectively. We utilize genome mining approaches to show how the genomes of these strains harbor more non-ribosomally synthesized peptide (NRP) clusters compared with other publically available B. licheniformis strains with complete genome sequences. Specifically, two NRP clusters are detected only in Bac48 and Bac84 with 52 and 46 genes respectively. The clusters are predicted to produce Fengycin with an average cumulative blast score of 31586 and Bacitracin with an average cumulative blast score of 45477 against the Minimum Information about a Biosynthetic Gene cluster database (MIBiG). We accordingly hypothesize that Bac48 and Bac84, two Red Sea strains, are good candidates to be used as microbial chassis for antimicrobial production especially lipopeptides.

ABSTRACT. Stem cell behaviors such as self-renewal and differentiation result from complex interactions among signaling, gene regulation, metabolism, and other pathways. Extensive research has elucidated many details of these pathways. However, we do not have a predictive understanding of how these pathways collectively determine behavior. To gain an integrated understanding of stem cells, we are developing a whole-cell computational model of the relatively well-characterized H1 hESC line.

Our model will be composed of multiple submodels of individual pathways such as signaling, gene regulation, and metabolism. The model will be based on a wide range of genomic and biochemical data. (1) We constructed an H1 genome by mapping published H1 WGS reads onto the hg19 reference. This genome will provide the foundation for the genes and reactions represented by the model. (2) We developed a metabolism submodel by (a) reconstructing the cell and media composition, protein expression, kinetic parameters, reaction fluxes, and growth rate of H1 hESCs; (b) constructing a seed network from the Recon 2.2 consensus model based on the reconstructed proteome; (c) parsimoniously add additional reactions from Recon 2.2 using CORDA2 until the network could produce each reconstructed cell component; (d) bounding the reaction fluxes by Michaelis-Menten rate laws; and (e) calibrating the submodel using the reconstructed cell composition, reaction fluxes, and growth rate. (3) We plan to (a) create submodels of signaling, gene regulation, RNA and protein synthesis and degradation, replication, and cell division; (b) expand the metabolic submodel with missing reactions that produce the metabolites required in other submodels; (c) integrate all submodels; and (d) validate the integrated model by comparison to observed hESC phenotypes such as its short G1 phase and rapid growth.

Ultimately, we aim to use the model to help elucidate the mechanisms of stem cell behaviors and help engineer stem cells for regenerative medicine.

ABSTRACT. An increase in the number of phosphoproteomic datasets obtained via high-throughput mass-spectrometric methods incites the need for tools that enable exploring them.

Minardo is a strategy to enable exploration of time-series phosphoproteomic dataset in terms of the corresponding kinases and phosphatases for observed phosphorylation alterations, overlaid on a cellular topology and depicted with time. The Minardo strategy has been successfully applied, by manually putting together the knowledge, to two high-throughput phosphoproteomic datasets. Here, we present work towards automation of Minardo, constructed as a web based tool.

The network identification methods (i.e. determining the involved kinases and phosphatase) combine multiple data-sources containing exact phosphatases and kinases for any given site (e.g. PhosphositePlus and DEPOD) as well as resources that allow us to hypothesise the possible enzymes that may be involved (such as BioGrid and Gene Ontology Annotations). Localisation datasets (such as Compartments) are also utilised to hypothesise the location of the occurrence of the phosphorylation or dephosphorylation reaction.

We provide an evaluation of the results obtained via our automated methods, for the two phosphoproteomic datasets, against their manually annotated versions. We observe good precision and poor recall for the kinase and phosphatase annotations, with phosphatase annotations having much worse recall.

These works represent a starting point for building tools for exploring phosphoproteomic datasets in the context of networks representing reactions at the level of sites. Our evaluation results of good precision and poor recall reflect on the current status of knowledge in existence for phosphorylation sites. We also provide a discussion of complementary methods, as well as future directions.

ABSTRACT. T helper (Th) 22 cells represent the newest member of the T helper family. As part of adaptive immunity, the physiological role of Th22 cells is to maintain the integrity of epidermal barriers. However, if this per se positive function is outbalanced it can turn pathologic and lead to deviations in the skin like they are seen in e.g. psoriasis. Aim of this project is to comprehensively characterize the Th22 phenotype and to identify similarities, differences and possible relationships to other known T helper subsets. To achieve this aim, we analyzed in total 60 T cell clones from different subsets derived from human biopsies or blood of chronic inflammatory skin diseases on protein and mRNA level by using Bioplex assays and whole transcriptome expression arrays, respectively, in a matched analysis setting. By using a combination of six clustering methods, the clones were grouped into subsets based on their cytokine secretion profile delivering several Th22 clones but also all other known Th subsets. These identified clusters were then compared using the transcriptomics data to describe similarities and differences of Th22 cells to other subsets. To obtain candidate genes, we fit several regularized logistic regressions (elastic net penalty) with the subset membership as outcome and all measured mRNA transcript as explanatory variables. We controlled the false discovery rate of the resulting candidate genes using stability selection. We found at least six genes, which are unique for the Th22 phenotype. Investigating the shared features revealed a higher similarity between Th22 and Th17 cells as compared to the other subsets. The selected genes are currently verified on the epigenetic level using DNA methylation data as well as experimentally for their role in Th22 differentiation. Specific targeting of Th22 cells by the identified genes may be useful to equilibrate outbalanced immune responses in human diseases.

ABSTRACT.     DNA exists as single molecules in individual cells. Consequently, gene expression is stochastic. We have recently developed a method for single cell transcriptome with high detection efficiency and accuracy, revealing intrinsic correlations among different genes, which has previously been masked within the stochastic noise.  For a particular human cell type, we were able to extract ~120 transcriptionally correlated modules acting from the stochastic gene expression data of ~700 individual cells under a non-equilibrium steady state condition. 

    The fact that there are 46 different individual DNA molecules (chromosomes) in a human cell dictates that genomic variations occur stochastically and cannot be synchronized among individual cells. Probing such genomic variations requires single-cell and single-molecule measurements. However, existing whole-genome amplification (WGA) methods are limited by low accuracy of copy-number variation (CNV) detection and low amplification fidelity. We have developed transposase-based methods for single cell whole genome amplification, which have superseded previous methods.  With the improved genome coverage, we developed a high resolution single cell chromatin conformation capture method, which allows for the first 3D genome map of a human diploid cell. Applications and implications to biology and medicine of these new approaches will be discussed.