With New Technique, Biologists Capture Molecules in Motion

Today’s images of biological molecules, including the COVID-19 virus, are so sharp that individual atoms are just about visible. To construct these kinds of three-dimensional images, snapshots of tens and often hundreds of thousands of molecules trapped in a thin layer of ice are collected with an electron microscope and then combined into a single, sharper, three-dimensional image by a computer. 

The technique is called single-particle cryo-electron microscopy, and in 2017, Joachim Frank, PhD, professor of biochemistry & molecular biophysics at Columbia University Vagelos College of Physicians and Surgeons, shared the Nobel Prize in Chemistry for developing the technique.

But biologists are also intensely interested in movies of molecules, because they provide clues about the way molecules carry out their functions—and how those functions could be altered with drugs. Cryo-EM is powerful, but it can only capture molecules in a single, static conformation.

A group of collaborators from Columbia, Wisconsin-Milwaukee, CUNY, and Arizona State University is now creating videos of molecules in motion with a new technique that extracts even more information hidden in the flood of data from the electron microscope.

The technique exploits the way molecules wiggle randomly in solution before they are flash-frozen. Data analysis using machine learning can uncover the important molecular motions from random snapshots captured by the electron microscope. This analysis, employing the mathematical technique of geometric machine learning, and refined using molecular dynamics simulations, uncovers the way molecules perform their biological function. 

Developed through a collaboration among the laboratories of Joachim Frank at Vagelos College of Physicians and Surgeons, Abbas Ourmazd, and Peter Schwander at the University of Wisconsin in Milwaukee, Amedee des Georges at CUNY, and Abhishek Singharoy at Arizona State University, the results were published this month in Nature Communications with joint first authors Ali Dashti and Ghoncheh Mashayekhi.

The molecule that served to demonstrate this novel technique of data analysis is the ryanodine receptor, which was previously solved at 3.4 Angstrom resolution by cryo-EM through the combined efforts of Frank, Andrew Marks, and Wayne Hendrickson at Columbia University Vagelos College of Physicians and Surgeons. When activated, this molecule releases calcium and triggers muscle contraction in skeletal muscle and in the muscles of the heart.  

By combining information from 800,000 images of the receptor with and without activating molecules present, the functional pathway between closed and open receptors was revealed for the first time, which may help in the development of new drugs for heart failure and other diseases.