By refining an emerging imaging technique and harnessing a powerful computer network, researchers have glimpsed the three-dimensional structure of a virus in unprecedented detail. The images, which capture the virus in a near-native state, have fine enough resolution that the backbones of individual proteins can be traced–a feat never before accomplished for a whole, intact organism.
“This particular achievement is basically a showcase of how this technique has progressed to a level that is near atomic resolution, and that allows us to image a real living infectious virus in a near-native environment,” says Wen Jiang, assistant professor of biological sciences at Purdue University, who led the study. The results appeared in the February 28 issue of the journal Nature.
The technique, called single-particle electron cryomicroscopy, involves flash-freezing whole viruses in a watery solution, a method that preserves their natural structure. The freezing process is so fast that the resulting ice is amorphous rather than crystalline, meaning that there are no ice crystals to damage the viral particles. The frozen sample is then bombarded with a beam of electrons from an electron microscope.
See other pictures of the virus.
Watch a movie of the virus model.
The resulting images allowed the researchers to distinguish between structures as close together as 4.5 angstroms. At just under half a nanometer, that distance is tiny enough to reveal near-atomic-level details. Earlier images of the same virus, also generated by Jiang’s group, yielded a resolution of about 9.5 angstroms–fine enough to illuminate some large-scale features of individual proteins, but not fine enough to trace those proteins’ backbones.
In other types of imaging, the structures to be imaged–whether they’re whole viruses or elaborate protein assemblies–must be arranged so that all the individual particles are oriented in exactly the same direction. That process requires advanced crystallization techniques to which many structures, including the virus used in this study, are not amenable.
The virus, called epsilon 15, belongs to a family of viruses that infect bacteria and have double-stranded DNA genomes and tails. “When you have a tail sticking out, it makes crystallization pretty hard,” says Jiang.
Single-particle electron cryomicroscopy eliminates the crystallization step entirely. The many particles to be imaged are arrayed and imaged in random orientations, and then a single composite three-dimensional model is assembled from tens of thousands of those images. In fact, the more orientations are captured, the better the model will be.
“It gives you the opportunity to solve structures that cannot be crystallized,” says Matsudaira.
The single-particle approach also improves image quality by protecting the relatively frail viruses from significant degradation by the microscope’s electron beam. Since the final model is constructed from images of a large number of viral particles, no single particle must be bombarded long enough to accumulate major damage.
Assembling the many images into a composite requires an enormous amount of computing power. Jiang and his group harnessed Purdue’s Condor pool, a resource that uses idle CPU cycles from computers across the campus. In total, the project took about one million CPU hours, spread over about 100 days.
It’s this computing power, along with refinements to the image-processing software, that allowed the researchers to generate such a high-resolution model. Because they had the resources to handle a massive input of data, they could combine many more images to create the composite.
Earlier uses of electron cryomicroscopy to model the structures of viruses have relied on shortcuts, such as assuming that the virus’s structure will be highly symmetrical. Thanks to the Condor pool, Jiang’s group was able to avoid such simplifications in determining how the virus’s surface proteins fit together.
“They did the pure experiment, which was to solve the structure without assuming symmetry,” says Matsudaira. That, he says, is the project’s most significant innovation–even more so than the 4.5-angstrom resolution.
From approximately 36,000 single-particle images, the researchers pieced together a model of epsilon 15’s protein shell, known as a capsid. Earlier work suggested that the capsid only incorporated one major protein. But in addition to tracing that protein’s backbone, the new model revealed a mysterious second protein–much smaller than the first–that no previous structural or biochemical study had predicted.
When the group reanalyzed the virus’s constituent proteins using a more sensitive biochemical screening method, they indeed found evidence of the smaller protein.
Jiang says that this outcome turns conventional structural analysis on its head. Usually, a particle’s biochemical makeup is called upon to help derive its structure. Here, the virus’s structure, as revealed by this powerful new analysis, helped uncover a previously overlooked biochemical feature.
“Usually the structure relies on the biochemistry,” says Matsudaira, “but this was exactly the opposite.”
In the future, Jiang hopes to further improve the resolution of images produced by single-particle electron cryomicroscopy. By further refining the software and perhaps invoking even more computing power, he anticipates that it may be possible to reach three-angstrom resolution within the next few years. That level of detail would uncover atomic-level features.
Beyond epsilon 15, the technique could be used to create structural models of other, more clinically relevant viruses. Jiang’s lab is currently applying the new approach to West Nile virus and dengue virus. Elaborate protein structures other than viral capsids would also be ideal targets.
“This is just scratching the surface of this technique,” says Jiang. “The potential of the technique is so much more than what we have achieved so far.”
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Everything you need to know about artificial wombs
Artificial wombs are nearing human trials. But the goal is to save the littlest preemies, not replace the uterus.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
Data analytics reveal real business value
Sophisticated analytics tools mine insights from data, optimizing operational processes across the enterprise.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.