 |
|
Research projectsThe Le Novère group’s research interests revolve around signal transduction in neurons, ranging from the molecular structure of proteins involved in neurotransmission to signalling pathways and electrophysiology. In particular, we focus on the molecular and cellular basis of neuroadaptation in neurons of the basal ganglia. By building detailed and realistic computational models, we try to understand how neurotransmitter-receptor movement, clustering and activity, influence synaptic signalling. Downstream from the transduction machinery, we build quantitative models of the integration of signalling pathways known to mediate the effects of neurotransmitters, neuromodulators and drugs of abuse. We are particularly interested in understanding the processes of cooperativity, pathway switch and bistability. The group provides community services that facilitate research in computational systems biology. In particular, we are leading the efforts in encoding and annotating kinetic models in chemistry and cellular biology, including the creation of standard representations, the production of databases and software development. The Systems Biology Markup Language (SBML) is designed to facilitate the exchange of biological models between different types of software. The Systems Biology Graphical Notation (SBGN) is an effort to develop a common visual notation for biochemists and modellers. The Simulation Experiment Description Markup Language (SED-ML) is designed to facilitate the reproduction of simulation procedures and the presentation of results. Moving from the form to the content, we are also developing standards for model curation (MIRIAM, MIASE), a format for describing simulation experiments (SED-ML), and controlled vocabularies (the Systems Biology Ontology, the Kinetic Simulation Algorithm Ontology, the TErminology for the Description of DYnamics) to improve the models. Finally, a model is only useful if it can be easily accessed and reused. BioModels Database is now the reference resource where scientists can store, search and retrieve published mathematical models of biological interest. Computational systems biology of dendritic spine signallingThe glutamatergic synapse is one of the main cellular components of the mammal brain, responsible for most of the cognitive processing and also for learning and memory. It is located on a specific portion of the neuron, the dendritic spine. The spine can be seen as an independent electrical and biochemical compartment, and thus as a unit of signal treatment and integration. The glutamatergic synapse is a very complex structure. The neurotransmitter receptors are embedded in complex multi-molecular assemblies, encompassing proteins of the pre- and postsynaptic sides. Glutamate released by the presynaptic terminals activates glutamate receptors of the AMPA type, which trigger the electrical response. This electrical response in turn allows the opening of glutamate receptors of the NMDA type. Those receptors let calcium flow in the spine, which results in the activation of many signalling cascades, leading for instance to synaptic plasticity or spine remodelling. The research projects of the team are centred on various components of the signal treatment in the spine of a particular neuron, the medium-spiny neuron of the striatum, involved in the control of voluntary motion and processes of reward. Modeling the function of calcium sensors involved in synaptic plasticity[Benedetta F Baldi, Stuart J. Edelstein, Massimo Lai; Alumni: Lu Li, David Marshall, Melanie I. Stefan]Learning processes are thought to rely on modification of synaptic activity such as long-term potentiation (LTP) and depression (LTD). The key event regulating these processes is calcium influx through the NMDA receptor (NMDAR). In the cell, this calcium influx affects many signalling cascades, in particular through the activation of calmodulin. We have built a full microscopic, kinetic model of calmodulin (Stefan et al. 2008) by extending the framework of concerted allosteric transitions (Stefan et al. 2009). This model provides an explanation of the fact that low concentrations of calcium-activated calcineurin trigger LTD and high concentrations of calcium-activated calcium/calmodulin-dependent protein kinase II (CaMKII) trigger LTP, while in both cases, the effect is mediated by the activation of calmodulin. We also showed that physiological concentrations of calmodulin cause a phenomenon of "ligand-depletion" affecting the dynamic range and cooperativity of the calcium-sensing (Edelstein et al. 2010) In order to study the role of calcium dynamics we systematically studied the effect of calcium inputs' amplitude, duration and frequency. We showed that the three characteristics affect the relative activation of calcineurin and CaMKII, but in different manner. CaMKII is a dodecameric protein that phosphorylates a wide range of targets, including itself and the glutamate receptors. Each monomer can exist in many different states, and the enumeration of all the possible combinations is infeasible. In order to relate the structure of the enzyme to its function as a molecular memory device, we created molecular models of the complex between calcium–calmodulin and CaMKII, and models of the phosphorylated forms of the kinase. Molecular dynamic simulations were used to generate alternative structures. To understand its allosteric properties, we developed highly detailed stochastic models of the function of CaMKII and its associated proteins, such as calmodulin and NMDAR.
Modelling glutamate receptors machinery at the mesoscopic level[Massimo Lai, alumni: Dominic Tolle, Ranjitta Dutta-Roy]It is likely that the supra-macromolecular structure of the postsynaptic membrane strongly influence signal transduction. The neurotransmitter receptors are embedded in complex multimolecular assemblies, encompassing proteins of the pre- and postsynaptic sides, but are also part of the cytoskeleton and the basal lamina. Moreover, the whole structure is dynamic and evolves, for example under the control of the neuronal activity. The position and movements of neurotransmitter receptors in and around synapses influences neuronal signal processing. Moreover, it has long been know that LTP is, in part at least, due to the appearance of new neurotransmitter receptors at the postsynaptic site. We used particle-based stochastic simulations to show that thermal diffusion alone can account for the incorporation of receptors at the synaptic specialisation within the timeframe of LTP expression (Tolle and Le Novè 2010a, 2010b). Our model predicts how the system behaves under various conditions affecting the free diffusion of receptors in the membrane, such as a change in biophysical parameter values, varying spatial parameters or quantity of interacting components. Receptors accumulate rapidly at the postsynaptic density under a number of biologically observed conditions. This accumulation is controlled by the number of scaffolding proteins and their affinity for the receptors but not the characteristics of the membrane or the initial location. In order to incorporate realistic behaviour of the glutamate receptors we developed allosteric models of AMPAR explaining the observed multiple conductance states.
Modelling signalling pathways involved in the plasticity of striatal neurons[Christine Hoyer; Alumni: Éric Fernandez, Noriko Hiroi, Lu Li, Renaud Schiappa] The projecting neurons of the striatum provide a crucial route for information transfer in the basal ganglia, involved in motor, psycho-motor and behavioural functions. Dopamine modulates the inputs coming from cortical glutamatergic terminals, providing a measure of the internal (hedonic) state. A protein phosphatase inhibitor, DARPP-32, has been identified as a major target for both dopamine and glutamate signalling. We constructed quantitative models of the signalling pathways known to mediate the effects of neurotransmitters, neuromodulators and drugs of abuse leading to phosphorylation and dephosphorylation of DARPP-32 (Fernandez et al. 2006, Le Novè et al 2008). The models are based on literature mining and original experimental data. Emphasis is put on the modulation by dopamine of glutamatergic signals.
Integration of biochemical and electrical models of the striatal medium-spiny neuron[Michele Mattioni]The function of a neuron can only be understood by taking into account the influence of signalling pathways on its electrical behaviour. The MSN has a large spiny dendritic tree that receives many different Neostriatum neurotransmitter (e.g. GABA, Glu), neuromodulators (e.g. DA, ACh, NE, Anandamine) and neuropetides (e.g. enkephalin, dynorphin, somatostin, Substance P, neurotensin). This broad spectrum of signals is then processed and integrated, modulating the electrical behaviour of the cell. we are developing accurate multi-compartment models of the MSN that merges the electrical behaviour of each compartment with a biochemical representation of the intracellular signalling pathways. In order to better explore the simulations results we've developed NeuronVisio, a GTK2 Interface with 3D capabilities for the NEURON software. SynSys - Synaptic Systems[Nicolas Le Novère, Junmei Zhu, Massimo Lai]
SynSys is a large European consortium that aims at describing and computing the molecular and functional architecture of the synapse, in order to provide insight into its mechanisms of function in vitro and in vivo,thereby explaining the diseases associated with the malfunctioning of the synapse. Within the consortium, our main tasks are the reconstruction of the molecular influence network (logical model) behind the function of the synapse, and the quantitative modelling of synaptic release, dynamics of neurotransmitters, and calcium signalling. Computational Systems BiologyThe practice of systems biology relies on interfaces, and in particular interfaces between the entities we study, whether molecules, pathways or cells, or interfaces between tools. If these interfaces are to be generic enough to allow all users to leverage on existing toolkits, the existence of community-developed, well-supported standards is a fundamental requirement, in addition to open resources for tools and parts. Over the last decade or so, several efforts have been launched in this direction, addressing encoding formats, ontologies and databases. Some of these are now well-established in the field and play a significant role in increasing the size and the quality of quantitative models. More importantly, they have served as a catalyst to improve the collaborative nature of the computational systems biology community. Those efforts are now developed as part of the COMBINE community.
Standards of reporting (MIRIAM and MIASE)[Camille Laibe, Nick Juty; Alumni: Dagmar Waltemath]
Most published quantitative models in biology are lost for the community because they are insufficiently characterised, which prevents them from being reused. With today’s increased interest in detailed biochemical models, it was necessary to define a minimum quality standard for the encoding of those models. The Minimal Information Required in the Annotation of Models (MIRIAM) (Le Novère et al 2005) is a set of rules for curating quantitative models of biological systems. Their application enables users to search collections of curated models with precision, quickly identify the biological phenomena that a given curated model or model constituent represents, and facilitates model reuse, model composition into large subcellular models, and format conversion. An important part of the standard consists in controlled annotation of model components, based on Uniform Resource Identifiers (URIs). MIRIAM Registry is an online infrastructure created to enable interoperability of this annotation (Laibe and Le Novère 2007). The core of this resource is a catalogue of data types, whether controlled vocabularies or primary data resources, which provides the means to generate and resolve MIRIAM URIs. The use of MIRIAM annotations by the community is still growing, and software tools have been developed that uses URIs as a glue to merge models and to integrate other datasets. MIRIAM’s guidelines deal mostly with the structure of the models but in order to use the models to run simulations and obtain numerical results, one needs additional information. The Minimum Information About a Simulation Experiment (MIASE) is a fledging effort to agree upon a set of mandatory information to add to relevant publications. Both MIRIAM and MIASE are part of MIBBI (Taylor et al 2008), a more general effort to coordinate the development of reporting guidelines.
Ontologies for Systems Biology[Nick Juty, Camille Laibe, Anna Zhukova; Alumnus: Mélanie Courtot, Christian Knüpfer, Dagmar Waltemath]
Whilst many controlled vocabularies exist that can be directly used to relate quantitative models to biological knowledge, there was previously no classification of the concepts themselves used in quantitative modelling. One of the goals of the Systems Biology Ontology (SBO) is to facilitate the immediate identification of the relationship between a model component and the model structure (Le Novère et al., 2007). SBO is currently made up of six different vocabularies: 1) an ontology of entities which may participate in an interaction, a process or relationship of biological significance (for example: ‘enzyme’ and ‘ribonucleic acid’); 2) a taxonomy of the roles of reaction participants (e.g. ‘catalyst’, ‘competitive inhibitor’); 3) a controlled vocabulary for parameter roles in quantitative models (for instance: ‘forward unimolecular rate constant’ and ‘Michaelis constant’); 4) a list of modelling frameworks that specify how to interpret a mathematical expression (such as: ‘continuous framework’ or ‘discrete framework’); 5) a classification of mathematical expressions used in biochemical modelling (e.g. ‘mass action rate law’, ‘Henri-Michaelis-Menten rate law’); and 6) a catalogue of interactions (for example: ‘non-covalent binding’ and ‘transport reaction’). The annotation of quantitative model components with SBO terms adds a layer of semantics necessary to convert models between different formalisms, to link mathematical representations of biochemical models with graphical notations such as the Systems Biology Graphical Notation (see overleaf), or semantically enriched computing formats to represent biochemical knowledge such as BioPAX. To complete SBO, which is designed to enrich model descriptions, we are developing an ontology of simulation methods, KiSAO aimed to be used with SED-ML (see below), and an ontology to characterise numerical descriptions of dynamic behaviours TEDDY.
Formal languages to encode models and simulations[Sarah Keating, Nicolas Le Novère, Nicolas Rodriguez; Alumni: Anika Oellrich, Duncan Berenguier, Marine Dumousseau, Dagmar Waltemath]
The Systems Biology Markup Language (SBML) is an XML language designed to facilitate the exchange of biological models between different simulators (Hucka et al. 2003). SBML is now an established standard in the field of systems biology, and is supported by several EMBL-EBI resources such as Reactome, IntAct and BioModels Database. We are now working to develop the new generation of SBML. The field of computational systems biology is now so wide and diverse that a single language, supported by all tools, cannot cover every approach. SBML Level 3 is therefore be modular, with a mandatory core package and optional modules. The group is particularly working on packages to represent multi-component, multi-state species, qualitative models, space and geometry, and hierarchical modelling. We use our generic SBML editor (Rodriguez et al. 2007) as a benchmark to test possible packages and for various related projects of the group. We also provide software to convert to and from SBML. SBML support is facilitated by the existence of standard library, such as libSBML in C++ and JSBML in Java, but also higher level layers such as SBMLToolbox to support SBML in MatLab. While SBML encodes the mathematical structure of the models, it does not specify how to obtain numerical results from this description. Together with simulator developers, we are creating a complementary format, the Simulation Experiment Description Markup Language (SED-ML) (Köhn and Le Novère 2008). A SED-ML file defines which models to simulate, how to modify them, which simulation approach to apply, how to post-process the numerical results and how to report them.
Systems Biology Graphical Notation[Nicolas Le Novère; Alumni: Lu Li, Sarala Wimalaratne]
Standard graphical representations have played a crucial role in science and engineering throughout the last century. Without electrical diagrams, it is very likely that our industrial society would not have evolved at the same pace. Similarly, specialised notations such as the Feynman notation or the process flow diagrams were instrumental for the adoption of concepts in their fields. With the advent of systems biology, and more recently of synthetic biology, the need for precise and unambiguous graphical descriptions of biochemical processes has become more pressing. While some ideas have been advanced over the last decade, with a few detailed proposals, no actual community standard has emerged. We developed the Systems Biology Graphical Notation (SBGN) (Le Novère et al., 2009), a graphical representation crafted over several years by a community of biochemists, modellers and computer scientists. Three orthogonal and complementary languages have been created: the Process Descriptions, the Entity Relationships and the Activity Flows. These three idioms enable scientists to represent any network of biochemical interactions in a standardised way, which can then be interpreted unambiguously. The set of symbols used is limited and the grammar kept as simple as possible, to also allow its use in textbooks and education. Shared SBGN languages will foster efficient and accurate representation, storage, exchange and reuse of information on biological knowledge, e.g. signalling pathways, metabolic and gene regulatory networks, between the communities of biologists, theoreticians and computational biologists.
BioModels Database[Ishan Ajmera, Camille Laibe, Vijayalakshmi Chelliah, Lukas Endler, Gael Jalowicki, Nicolas Rodriguez, Michael Schubert; Alumni: Alexander Broicher, Marco Donizelli, Enuo He, Arnaud Henry, Ron Henkel, Chen Li, Kedar Nath Natarajan, Jean-Baptiste Pettit, Melanie I. Stefan, Karim Tazibt]
For computational modelling to become more widely used in biological research, modellers must be able to exchange and share their results. BioModels Database is a data resource that allows modellers to store, search and retrieve published mathematical models of biological interest (Li et al. 2010, Le Novère et al. 2006). Models are annotated and linked to other relevant data resources. BioModels Database accelerates computational modelling efforts by allowing researchers to leverage each others’ work more directly. It also supports improved and more accurate communication of research results by allowing journal publishers to encourage the submission of models in the same electronic format, stored in a common, publicly accessible, location. Finally, the database provides examples of working models for educational purposes, allowing inexperienced modellers to find ready-to-use models for exploration. BioModels Database has been developed in collaboration with the California Institute of Technology and is now the largest database of curated models worldwide (containing more than 470 models and 45 000 reactions). This status is recognised by BioMedCentral, Nature Publishing Group and the Public Library of Science, all of which request deposition of models upon submission of manuscripts to several hundreds of journals. We regularly release new versions of the database, with new features for both users and curators.
|
|
