Simulation of Electron Transfer with a Shape-Based Grained Multi-Agent Model
Résumé
The respiratory chain is a biological process which produces energy in the cell. This process relies on electron transfer between macromolecular proteic complexes located in the mitochondrion inner membrane. During this process, the elec- trons are driven from different sites inside the macromolecules. The redox state change of a site has a local effect on the atom positions and can cause a global change of the proteic structure: this is a conformational change. Such modification can affect the behavior of the protein and can disturb the whole process. In addition, proteins are mobile in the membrane and this special medium has serious constraints on molecule encounter which is necessary for electron to be transfered from one complex to another. Models based on differential equations have already been developed for the respiratory chain but they do not include spatial properties, such as localization or conformation of proteins. So, to be more realistic, models have to take into account the possibility of conformational changes and spatial localization of proteins. A spatial representation of proteins can be obtained by using atom positions and, thus, modeling movements of proteins is possible. Molecular dynamics is a simulation technique which use these data to simulate the movements of all atoms of molecules. Though it allows very realistic simulations, the systems that could be simulated are limited in their size and the total simulated time is short (a few nanoseconds). To simulate larger systems, the original molecular dynamic model has been adapted by reducing the number of objects to simulate. Specific models for proteins, like the Residue- Based Coarse-Grained (RBCG) or the Shape-Based Coarse-Grained (SBCG) models, have been designed to allow the simulation of multiproteic systems and the simulated time can reach the microsecond time scale. To model the respiratory chain at the molecular level, both spatial representation and enzymatic behavior are required. As differential equations nor MD simulations allows us to put together this two aspects, we consider to use a multi-agent system. Multi-agent systems are composed of multiple autonomous entities, the agents, in interaction. These agents can interact following a set of embedded rules and with the agent paradigm, it is possible to model a situated entity with interactions based on a physical model. To design the movement of proteins we have defined a model base on Shape-Based Coarse-Grained approach. A first application have been made including two complexes of the respiratory chain (complex II and complex III) and a generic lipid for membrane model. A specific reactive agent model includes the definition of a physical body composed of grains. A lot of simulations have shown that the electron chain can be influenced by movements of proteins and give explanations of several experimental results that are not easy to interpret directly from experiment. Next step will focus on the description of the two others complexes of the respiratory chain (complex I and complex IV) and will test the hypothesis of hyperstructure building.