Self-Driving Microscope Tracks and Analyzes Misfolded Protein Aggregation in Real Time

Yet to the human eye, proteins that are destined to form harmful aggregates appear identical to normal ones, and these aggregates form rapidly and randomly—within minutes. Detecting and understanding the formation of such aggregates is crucial, as their biomechanical properties are directly linked to disease progression and disruption of cellular function. However, imaging tools that rely on fluorescent labels may alter cell properties and hamper accurate analysis. Now, researchers have developed a real-time imaging system capable of tracking protein aggregation dynamically and even predicting its onset before it begins.

The self-driving imaging system, developed by researchers at EPFL (Lausanne, Switzerland), in collaboration with the European Molecular Biology Laboratory (Heidelberg, Germany), builds on previous work involving deep learning algorithms that could detect mature protein aggregates in unlabeled images of living cells. The team developed two distinct algorithms. The first is an image classification algorithm that activates a Brillouin microscope—normally too slow for live-cell imaging—only when it detects mature aggregates. Brillouin microscopy uses scattered light to characterize the biomechanical properties of aggregates, such as elasticity. The second algorithm is an “aggregation-onset” detection tool trained on fluorescently labelled images, capable of distinguishing subtle differences and predicting when aggregation will occur with 91% accuracy. This predictive function enables the microscope to be activated precisely when needed, capturing the biomechanics of protein aggregation as it unfolds.

The researchers tested and validated the system by observing the full dynamic formation of aggregates and measuring their properties in real time. Their findings, published in Nature Communications, demonstrated how self-driving microscopy could incorporate label-free methods for broader biological use. The ability to foresee and capture aggregation processes has significant implications for drug discovery and precision medicine, especially in targeting toxic oligomers suspected to drive neurodegeneration. Going forward, the researchers aim to develop drug discovery platforms based on this technology to accelerate the development of more effective therapies for neurodegenerative diseases.

“This is the first publication that shows the impressive potential for self-driving systems to incorporate label-free microscopy methods, which should allow more biologists to adopt rapidly evolving smart microscopy techniques,” said EPFL PhD graduate Khalid Ibrahim, who led the collaborative effort.

Related Links:
EPFL
European Molecular Biology Laboratory