Today’s digital photos are far more vivid than just a few years ago, thanks to a steady stream of advances in optics, detectors, and software. Similar advances have also improved the ability of machines called cryo-electron microscopes (cryo-EMs) to see the Lilliputian world of atoms and molecules. Now, researchers report that they’ve created the highest ever resolution cryo-EM image, revealing a druglike molecule bound to its protein target at near atomic resolution. The resolution is so sharp that it rivals images produced by x-ray crystallography, long the gold standard for mapping the atomic contours of proteins. This newfound success is likely to dramatically help drugmakers design novel medicines for a wide variety of conditions.
“This represents a new era in imaging of proteins in humans with immense implications for drug design,” says Francis Collins, who heads the U.S. National Institutes of Health in Bethesda, Maryland. Collins may be partial. He’s the boss of the team of researchers from the National Cancer Institute (NCI) and the National Heart, Lung, and Blood Institute that carried out the work. Still, others agree that the new work represents an important milestone. “It’s a major advance in the technology,” says Wah Chiu, a cryo-EM structural biologist at Baylor College of Medicine in Houston, Texas. “It shows [cryo-EM] technology is here.”
Cryo-EM has long seemed behind the times—an old hand tool compared with the modern power tools of structural biology. The two main power tools, x-ray crystallography and nuclear magnetic resonance (NMR) spectroscopy, enable researchers to pin down the position of protein features to less than 0.2 nanometers, good enough to see individual atoms. By contrast, cryo-EM has long been limited to a resolution of 0.5 nm or more.
Cryo-EM works by firing a beam of electrons at a thin film containing myriad copies of a protein that have been instantly frozen in place by plunging them in liquid nitrogen. Detectors track the manner in which electrons scatter off different atoms in the protein. When an image is taken, the proteins are strewn about in random orientations. So researchers use imaging software to do two things; first, they align their images of individual proteins into a common orientation. Then, they use the electron scattering data to reconstruct the most likely position of all the protein’s amino acids and—if possible—its atoms.
Cryo-EM has been around for decades. But until recently its resolution hasn’t even been close to crystallography and NMR. “We used to be called the field of blob-ology,” says Sriram Subramaniam, a cryo-EM structural biologist at NCI, who led the current project. But steady improvements to the electron beam generators, detectors, and imaging analysis software have slowly helped cryo-EM inch closer to the powerhouse techniques. Earlier this year, for example, two groups of researchers broke the 0.3-nm-resolution benchmark, enough to get a decent view of the side arms of two proteins’ individual amino acids. Still, plenty of detail in the images remained fuzzy.
For their current study, Subramaniam and his colleagues sought to refine their images of β-galactosidase, a protein they imaged last year at a resolution of 0.33 nm. The protein serves as a good test case, Subramaniam says, because researchers can compare their images to existing x-ray structures to check their accuracy. Subramaniam adds that the current advance was more a product of painstaking refinements to a variety of techniques—including protein purification procedures that ensure each protein copy is identical and software improvements that allow researchers to better align their images. Subramaniam and his colleagues used some 40,000 separate images to piece together the final shape of their molecule. They report online today in Science that these refinements allowed them to produce a cryo-EM image of β-galactosidase at a resolution of 0.22 nm, not quite sharp enough to see individual atoms, but clear enough to see water molecules that bind to the protein in spots critical to the function of the molecule.
That level of detail is equal to the resolution of many structures using x-ray crystallography, Chiu says. That’s vital, he adds, because for x-ray crystallography to work, researchers must produce millions of identical copies of a protein and then coax them to align in exactly the same orientation as they solidify into a crystal. But many proteins resist falling in line, making it impossible to determine their x-ray structure. NMR spectroscopy doesn’t require crystals, but it works only on small proteins. Cryo-EM represents the best of both worlds: It can work with massive proteins, but it doesn’t require crystals.
As a result, the new advances could help structural biologists map vast numbers of new proteins they’ve never mapped before, Chiu says. That, in turn, could help drug developers design novel drugs for a multitude of conditions associated with different proteins. But one thing the technique has already shown is crystal clear, that in imaging, as well as biology, slow, evolutionary advances over time can produce big results.