Integrative approaches for signalling and metabolic networks

Vassily Hatzimanikatis is Associate Professor of Chemical Engineering and Bioengineering at École Polytechnique Fédérale de Lausanne (EPFL). He received his Diploma in Chemical Engineering from the University of Patras, in Greece, and his MS and PhD in Chemical Engineering from the California Institute of Technology. He held academic positions at the Swiss Federal Institute of Technology (ETH) in Zurich, Switzerland and at Northwestern University in Evanston, Illinois, and he worked in industry (DuPont, Cargill, and Cargill Dow) on the development of biocatalysts for the production of industrial chemicals. Vassily’s research interests are in the areas of systems biotechnology, metabolism and metabolic engineering, bioenergetics and systems medicine. He is an Associate Editor of Integrative Biology. He has published over 80 technical articles and he is a co-inventor in three patents. His work has been recognized by the 2014 Metabolic Engineering Award from the International Society of Metabolic Engineering (IMES) and the 2011 Gaden Award of the ACS Biochemical Technology Division.

Julio Saez-Rodriguez is a Professor at the Joint Research Center for Computational Biomedicine in Aachen, Germany. He is also an affiliated member of Sage-Bionetworks and a director of the DREAM initiative to catalyze the development of methods in systems biology (http://www.dreamchallenges.org). He studied Chemical Engineering at the Universities of Oviedo and Stuttgart. He obtained his PhD at the University of Magdeburg in 2007 upon work conducted at the Max-Planck-Institute of Dynamics of Complex Technical Systems with E. D. Gilles. His PhD was awarded the MTZ Foundation Award 2008 for the best PhD thesis in medicine-oriented Systems Biology. He was a postdoctoral fellow at Harvard Medical School with Peter Sorger and Doug Lauffenburger at M.I.T., and a Scientific Coordinator of the NIH-NIGMS Cell Decision Process Center. From 2010 until 2015 he was a group leader at the European Bioinformatics Institute (EMBL-EBI) with joint appointment in the EMBL Genome Biology Unit, as well as a senior fellow at Wolfson College (Cambridge). His research is on the development and application of computational methods to acquire a functional understanding of signaling networks and their deregulation in disease, and to apply this knowledge to develop novel therapeutics.

The study and analysis of the organization of biochemical reactions into complex networks is central to integrative biology. This themed issue on networks has assembled reviews and original research papers that demonstrate how biological chemistry from the basic mechanistic design of individual reactions to hundreds of reactions organized into networks determine cellular function. The collection spans from metabolism to signal transduction, and from detailed biochemical formalisms to coarse graph-based approaches.

We can appreciate the model development process, the methods for the analysis and the applications of genome-scale models of metabolic networks through the thorough review of the genome-scale models of yeast by Sánchez and Nielsen (DOI: 10.1039/C5IB00083A). The authors present the history of yeast metabolic models and they discuss how omics data from different levels can be used in the development of such models and how in turn these models can be used for the integration, analysis and interpretation of different omics data. In a similar direction Sang Yup Lee and colleagues (DOI: 10.1039/C5IB00002E) offer a review of genome-scale human metabolic modeling, and they discuss how state-of-the-art high-throughput techniques and data are analyzed by advanced computational methods to formulate tissue/cell type-specific human metabolic models. Such context-specific metabolic models are used for studying metabolic diseases, the role of metabolism in other diseases such as cancer and infection, and the capabilities of biological chemistry to produce fuels and chemicals.

Mahadevan, Stephanopoulos and colleagues (DOI: 10.1039/C5IB00095E) assembled the biochemical reactions in the metabolic network of Moorella thermoacetica, a versatile acetogenic bacterium that is able to fix CO₂ and transform syngas (CO + H₂) into acetyl-CoA. These biosynthetic capabilities make this organism an attractive host for many biotechnology applications. The authors demonstrate how stoichiometric modeling and flux balance analysis can help us understand better the chemistry that leads from carbon dioxide and carbon monoxide to life and to every-day useful chemicals. E. coli is the benchmark organism for the analysis and design of metabolism in the development of almost every industrial chemical. In another study that combined metabolic modeling and experiments, Mahadevan and colleagues (DOI: 10.1039/C5IB00096C) have investigated the role of redundant reductions in the physiology of E. coli through the analysis of double-knockout mutants. They found that these mutants are not agnostic to the order in which their genes are deleted, and the order in which genes are deleted determines the phenotype of the mutants during the sub-optimal growth phase. However, these observations could not be explained using stoichiometric models alone and they would require the consideration of kinetic models that describe the regulatory effects at the level of gene expression and the regulation of enzyme kinetics.

The integration of enzyme kinetics for the analysis of the dynamic properties of these networks is one of the major challenges in the analysis of large- to genome-scale metabolic networks. The contribution by Klipp et al. (DOI: 10.1039/C5IB00050E) proposes a method for contextualizing existing dynamic metabolic models through the formulation and addition of reactions that account in a simplified but consistent manner for some of the pathways that are missing from the dynamic metabolic model. Such approaches can help increase significantly the accuracy of dynamic metabolic models. The nonlinear enzyme kinetics in the formulation of dynamic metabolic models introduce many challenges common in nonlinear dynamics. One of these challenges is the robustness of the system, which is the ability of the system to maintain a stable, physiological steady state. Liao and colleagues (DOI: 10.1039/C4IB00257A) propose a metric that can allow characterization of the robustness of a nonlinear dynamic system. Such a metric can be used for the analysis and design of complex biochemical networks and for the improvement of mathematical models that describe these systems.

Gunawardena and colleagues (DOI: 10.1039/C5IB00009B) further study the issue of robustness in the specific context of bifunctional enzymes. These enzymes control robustly different biochemical processes such as metabolic branch points and osmoregulatory networks. The authors use tools of computational algebraic geometry to find relationships (invariants) between the steady state concentrations of the modified and unmodified substrate. This allows them to characterize robust behavior in these systems, and illustrate the power of these invariants to study biochemical networks.

Similarly to metabolism, formulating and analyzing models that describe the underlying biochemical reactions is the most common way to study signaling networks. Analytical methods that rely on the biochemical structure, such as those presented by Gunawardena et al., can be applied to identify key properties. Since signaling is largely a dynamic process, the reactions are often converted into differential equations for simulation and analysis. However, the size and complexity of the models increase exponentially with the scope of the model, and require parameter values often unknown.

As an alternative, qualitative models and in particular logic-based models have become popular recently to analyze properties of large signaling networks. In this issue, Zinovyev and colleagues (DOI: 10.1039/C5IB00029G) use Boolean formalism (the simplest variant of a logic model, where each variable is either ON or OFF) to characterize genetic interactions (i.e. epistasis) in signaling and regulatory networks. “Genetic interactions” refers here to cases where the combined effect of two genes cannot be predicted from the effect of both of them alone. Genetic interactions are useful to characterize functional relationships between genes, and can provide novel insight to treat diseases. Because Boolean logic models are so simple, they can be built for many systems, even if our knowledge is not very refined, and hence their approach can be broadly applicable.

While the knowledge required to build logic models is significantly less than for biochemical models, there are many pathways for which there is information about the proteins involved and the potential connections (directed and often signed), but not enough granularity to build logic functions. In addition, while fairly efficient and scalable, the methods to simulate and analyze these models can generally not handle the size needed to ultimately cover genome-wide networks. Melas et al. (DOI: 10.1039/C4IB00294F) propose an approach to identify, in directed and signed networks of thousands of proteins, the pathways that causally link a given perturbation with an altered profile of gene expression. The authors apply their method to study the mode of action of drugs in the context of drug induced lung injury, but it can also be used to dissect the effect of other perturbations or alterations (e.g. mutations) on downstream processes.

Through the articles described above, this collection provides a glance at the variety and power of network biology approaches. We hope to see more papers on this area in the future in Integrative Biology. An area of active research is around models that bridge the different layers of molecular processes within cells (Gonçalves et al., Molecular BioSystems, 2013, 9, 1576, DOI: 10.1039/C3MB25489E).

We can foresee further development of the computational methods, refinement of the models and integration of additional accumulated data from multiple levels of the genome sequence to metabolite levels. Thereby these models should become increasingly useful across many areas of fundamental and integrative biology, as well as in applications ranging from biotechnology to analysis of disease and eventually the design of personalized therapies.

---