"Our simulation on the CSC supercomputer consisted of about 120000 atoms” says Lindahl.

The pilot Grand Challenge research project, undertaken by a group led by Professor Ilpo Vattulainen (Tampere University of Technology) and Assistant professor Erik Lindahl (Stockholm University) was run on CSC's new supercomputer, “Louhi”, in Spring of 2007. The research consisted in simulating the dynamics of a protein – an ion channel – that plays a key role in many biological processes.

Ion channels: a hot topic

An ion channel is a protein that is present in the membrane that surrounds all biological cells. It is mainly responsible for transporting small charged particles – ions – a cross the otherwise impermeable membrane. “In general, ions cannot pass through cell membranes unaided. The protein provides a passage-way through the membrane”, states E. Lindahl, one of the leaders of the project.

Side-view of the protein. The protein is shown in blue. The water molecules are shown in red and white and the potassium ions in yellow. The lipid bilayer has been cut away for clarification.

“For the project, we focused on one specific type of ion channels, namely “voltage-gated” ion channels, in which the opening of this passage is normally blocked by a part of the protein. This forms a “gate”, which can open when the electrical potential over the membrane changes”, he explains. “These channels are involved in key biological processes, such as the neuronal action potential, the coordination of the heart beat and the contraction of skeletal muscles, among others. Studying them is thus a matter of real importance”, he concludes.

Ion channels are indeed one of today´s hottest research topics, due not only to their potential uses but also to the relatively limited knowledge that we have of them. The structure of these membrane proteins and, in particular, the mechanisms associated with their structural transition – the process by which they “open” and “close” – are not particularly well understood. “Although more than 100,000 structures of soluble proteins are already known, only around 100 membrane protein structures have been experimentally determined, despite the fact that membrane proteins account for about one third of all the proteins in a cell”, notes Pär Bjelkmar, one of Lindahl´s Ph.D. students, who took an active part in the project.


Side-view of the ion channel system. The membrane is visible in pale blue. The embedded protein is shown in green and the water molecules on either side of the membrane are shown in red and white.

Efforts to understand the structure of these membrane proteins won Roderick MacKinnon the Nobel Prize for Chemistry in 2003. “Since then, these proteins have become a model system for studying the mechanics of membrane proteins as they undergo large structural changes. Because these proteins are of such fundamental importance for the workings of cells, they have been studied extensively and their function is hotly debated within the scientific community”, Bjelkmar explains.

From real experiments to computer simulations

While it is possible to retrieve the exact positions of all the atoms in an ion channel experimentally, this generally only provides a static picture of the proteins, not capturing their dynamic elements of the system. It is here that computational modeling can add considerable value to experimental work.

“The beauty with computer simulations is that they are an extremely good complement to “real” experiments. They provide an extremely detailed picture of the intricate workings of the system studied on an atomistic level. Starting from a static description of the system, typically given by various experimental data, the computer calculates the motions of all atoms in the system according to the laws of physics. One could say that the computer experiment results in a movie, or “trajectory” as it is called, that depicts the exact behavior of the system”, explains Bjelkmar.

Simulations do require, however, very high computation capacities, particularly in the case of large proteins such as ion channels: “Our simulation on the CSC supercomputer consisted of about 120000 atoms” says Lindahl. “For each frame in the trajectory, the positions of all of these atoms have to be calculated. In addition, because atoms move so fast, we have to recalculate their positions very frequently. Thus, in order to simulate biologically relevant time-scales (i.e. long enough “movies”), supercomputers are very much required”, adds Bjelkmar.

The project carried on CSC supercomputer Louhi has been a complete success: “With the help of the supercomputer Louhi and new state-of-the-art software, we have performed the longest-ever simulation on a comparable system”, says Lindahl. “This simulation enabled us to show structural changes within the protein and membrane that occur within time-scales that previous efforts to simulate these systems could not capture. For example, we have seen large rotations of certain important structural elements within the protein, which we think is indicative of an initiation of the channel beginning to close”, explains Lindahl.

“We have also been able to describe the interactions of the protein and the membrane; the lipid molecules in the membrane have very different properties depending on their location relative to the protein. It seems the lipids close to the protein form a structural extension of that protein, and their special behavior is very possibly important for the binding of external effector molecules, such as drugs and toxins”, adds Bjelkmar. “Our simulation will provide new insights into the understanding of the structural transitions associated with the opening and closing of the channel, a topic hotly debated within the scientific community”, concludes Lindahl. ■

Damien Lecarpentier