Researchers from the University of Copenhagen have developed a machine learning algorithm that can track protein clumps under a microscope in real time, a groundbreaking advance in the study of neurodegenerative diseases such as Alzheimer's and Parkinson's. The algorithm automates the mapping and tracking of these clumps, a previously time-consuming task, and could speed up the development of new treatments for these diseases. Credit: SciTechDaily.com
Protein aggregation underlies many neurodegenerative diseases that affect the brain. Alzheimer's disease The risk of developing dementia is caused by the accumulation and loss of proteins. Scientists at the University of Copenhagen have developed a new instrument to detect and analyse these small protein clusters. This breakthrough provides deeper insight into the body's basic building blocks, potentially improving therapies for diseases such as cancer, Alzheimer's and Parkinson's.
Around 100,000 Danish people over the age of 65, and more than 55 million people worldwide, suffer from dementia-related diseases such as Alzheimer's and Parkinson's. These diseases occur when small cellular components clump together and impair important bodily functions. Scientists have yet to understand why they occur and how to treat them. Until recently, the lack of suitable tools made investigating these diseases extremely difficult.
Now, researchers from the Hatzakis Laboratory in the Department of Chemistry at the University of Copenhagen have Machine Learning An algorithm that can track aggregation under a microscope in real time. The algorithm can automatically map and track key features of the aggregated building blocks that cause Alzheimer's and other neurodegenerative diseases. Until now, this has not been possible.
“Our algorithm solves a problem in just minutes that would take researchers weeks to solve. We hope that making it easier to study microscopic images of aggregated proteins will deepen our knowledge and, in the long term, lead to new treatments for neurodegenerative brain diseases,” said Dr. Jakob Kester-Hansen from the Department of Chemistry, who led the algorithm's research team together with Nikos Hazakis.
The study was published in a prestigious scientific journal. Nature Communications.
Instant detection of minute proteins
Proteins and other molecules coming together to exchange compounds and signals happens billions of times in our cells as a natural process that allows our bodies to function. But when errors occur, proteins can aggregate and interfere with their ability to function as intended. This can lead to neurodegenerative diseases of the brain and cancer.
The researchers' machine learning algorithm can spot protein clumps a billionth of a meter in microscopic images. At the same time, the algorithm can count the clumps, group them according to their shape and size, and track their development over time. The appearance of the clumps can have a big impact, for better or worse, on their function and how they behave in the body.
Insulin protein clumping together. Photo by Jacob Köstel-Hansen.
“When you look at the clumps under a microscope, you can quickly see that some are round and some have a thread-like structure, for example, and their exact shape can vary depending on the disturbances they cause. But sitting down to count them manually thousands of times would take a very long time that would be better spent doing other things,” says first author Steen Bender from the Department of Chemistry.
In the future, this algorithm will make it much easier to learn more about why clumps form, allowing new drugs and treatments to be developed to combat these disorders.
“Fundamentally understanding these masses depends on being able to observe them, track them, quantify them and describe what they look like over time, and currently there is no other way to do this automatically and effectively,” he says.
The tool is free for everyone to use
Researchers from the Department of Chemistry are now conducting full-scale experiments with this tool. Insulin When insulin molecules clump together, they become less able to regulate blood sugar levels.
“We also see this unwanted aggregation in insulin molecules. Our new tool allows us to see how added compounds affect these aggregations. In this way, the model helps us understand how to prevent aggregation or turn it into less dangerous or more stable aggregations,” explains Jacob Köstel-Hansen.
The researchers therefore believe that, once the microscopic building blocks are clearly identified, it is very possible that this tool could be used to develop new medicines. They hope that their work will spark a more comprehensive collection of knowledge about the shape and function of proteins and molecules.
“As other researchers around the world start to deploy this tool, it will help us create a large library of molecular and protein structures relevant to different diseases and biology in general, which will allow us to better understand and thwart diseases,” concludes Nikos Hatsozakis from the Department of Chemistry.
The algorithm is open source and freely available on the internet, making it available for use by scientific researchers and anyone interested in understanding aggregation of proteins and other molecules.
Reference: “SEMORE: Automating Super-Resolution Data Analysis with Machine Learning Segmentation and Morphological Fingerprint Recognition,” Steen WB Bender, Marcus W. Dreisler, Min Zhang, Jacob Kæstel-Hansen, Nikos S. Hatzakis, 26 February 2024, Nature Communications.
Publication date: 10.1038/s41467-024-46106-0
The research was supported by the Novo Nordisk Foundation's Centre for Optimised Oligo Evasion and Disease Control and was carried out by Steen W. Bender, Marcus W. Dreisler, Min Zhang, Jacob Kæstel-Hansen and Nikos S. Hatzakis from the Department of Chemistry.
