The ever-changing shape of the interior of cells limits our understanding to whatever snapshot we’re looking at under the microscope. Now, a new computer model developed at the University of Copenhagen is allowing researchers to predict how the most dynamic elements in a cell will develop. The advancement could play a role in helping to understand Alzheimer's, Parkinson's, cancer and other diseases in the future.
Cells are the smallest living components in all organisms. In a gigantic collaboration, trillions of them ensure that all biological processes in the human body operate as they should. For each cell to function properly, its various functions are distributed in small molecular compartments.
But because many of these compartments are constantly changing shape, it is difficult to capture anything but momentary snapshots of the life going on within cells. This makes it difficult to determine whether a particular cell will develop defects that can lead to diseases such as cancer, Alzheimer's and Parkinson's.
Now, researchers at the Linderstrøm-Lang Centre for Protein Science at the University of Copenhagen’s Department of Biology have developed a new computer model that can help researchers better understand and predict how cells organize. The model can simulate how proteins within cells collaborate on vital tasks.
"For more than a hundred years, we have known that human cells are highly organized and that different biological functions are found in various cell compartments. Over the past 10 years, we have learned about a new type of organization in cells that is entirely distinct from the structures that we were familiar with before. However, as these newly discovered structures are quite dynamic and constantly change their shape and composition, it’s difficult to understand how they’re connected and what governs their formation. Therefore, we developed an algorithm that can predict their properties," explains Professor Kresten Lindorff-Larsen.
A revolution in understanding cells
The tool is an advanced computer model based on biophysical measurements and a form of artificial intelligence that allows researchers to simulate various scenarios in which cellular molecules interact with each other. Kresten Lindorff-Larsen compares it to being somewhat like how weather is forecast for the weather apps we use.
"Our previous methods would be similar to being able to predict the weather a few minutes out. But instead of making very detailed predictions, we have built a model that provides us with the big picture and makes broader predictions," explains Kresten Lindorff-Larsen.
In particular, the researchers studied a process in cells where different types of molecules, like proteins and RNA, spontaneously gather to collaborate on a specific task.
"The cell is not just some bag in which things occur randomly. It is extremely organised and highly crowded. It collects many of its molecules in small droplets, in accordance with whatever process they are collaborating on. When done, they often separate again. This continues until a defect occurs that makes it no longer possible to perform the task correctly. Our new computer model can describe one of the processes underlying this organization in unprecedented detail," explains Postdoc Giulio Tesei.
Cancer, Alzheimer's and Parkinson's
Defects within cells manifest themselves in a wide range of diseases, including cancer, Alzheimer's and Parkinson's. Zooming in on the brain cell of a Parkinson's patient, one sees abnormal proteins clumping together in the form of elongated sticks. This is a process that many doctors and researchers have observed and studied previously, but which will now be easier to understand and perhaps eventually be treated.
"One goal with the model is for us to be able to, for example, go in and identify molecules and processes in a cell that are central to the formation of these sticks before they form. In doing so, one can better target the treatment before they begin aggregating inappropriately," explains Kresten Lindorff-Larsen.
The researchers emphasize that the model cannot stand alone, and must be used in conjunction with experiments. The research was published in the scientific journal PNAS and funded by the Lundbeck Foundation and the EU’s Horizon 2020 programme. Preprint of the study is available here.
- Together with graduate student Thea K. Schulze and Professor Ramon Crehuet from Barcelona, Professor Kresten Lindorff-Larsen and postdoctoral researcher Giulio Tesei have developed an algorithm to learn about the properties of molecules by conducting experimental studies on them.
- The model is based on a form of artificial intelligence in which a machine learns the rules of how molecules interact using biophysical experiments from known proteins.
- The researchers demonstrated that their model can accurately predict the propensity of different known proteins to come together in small droplets.