The new technique managed to generate structures from a mixture that contained all four of these organic compounds.

M. Martynowycz et al., ChemRxiv (2018), adapted by E. Petersen/Science (CC BY-NC-ND 4.0)

‘A new day for chemistry’: Molecular CT scan could dramatically speed drug discovery

In chemistry, structure rules because it determines how a molecule behaves. But the two standard ways to map the structure of small organic molecules, such as pharmaceuticals, hormones, and vitamins, have drawbacks. This week, two research teams report they’ve adapted a third technique, commonly used to chart much larger proteins, to determine the precise shape of small organic molecules. The new technique works with vanishingly small samples, is blazing fast, and is surprisingly easy.

“I am blown away by this,” says Carolyn Bertozzi, a chemist at Stanford University in Palo Alto, California. “The fact that you can get these structures from [a sample] a million times smaller than a speck of dust, that’s beautiful. It’s a new day for chemistry.”

The gold standard for determining chemical structures has long been x-ray crystallography. A beam of x-rays is fired at a pure crystal containing millions of copies of a molecule lined up in a single orientation. By tracking how the x-rays bounce off atoms in the crystal, researchers can work out the position of every atom in the molecule. Crystallography can pinpoint atomic positions down to less than 0.1 nanometers, about the size of a sulfur atom. But the technique works best with fairly large crystals, which can be hard to make. “The real lag time is just getting a crystal,” says Brian Stoltz, an organic chemist at the California Institute of Technology (Caltech) in Pasadena. “That can take weeks to months to years.”

The second approach, known as nuclear magnetic resonance (NMR) spectroscopy, doesn’t require crystals. It infers structures by perturbing the magnetic behavior of atoms in molecules and then tracking their behavior, which changes depending on their atomic neighbors. But NMR also requires a fair amount of starting material. And it’s indirect, which can lead to mapping mistakes with larger druglike molecules.

The new approach builds on a technique called electron diffraction, which sends an electron beam through a crystal and, as in x-ray crystallography, determines structure from diffraction patterns. It has been particularly useful in solving the structure of a class of proteins lodged in cell membranes. In this case, researchers first form tiny 2D sheetlike crystals of multiple copies of a protein wedged in a membrane.

But in many cases, efforts to grow the protein crystals go awry. Instead of getting single-membrane sheets, researchers end up with numerous sheets stacked atop one another, which can’t be analyzed by conventional electron diffraction. And the crystals can be too small for x-ray diffraction. “We didn’t know what to do with all these crystals,” says Tamir Gonen, an electron crystallography expert at the University of California, Los Angeles (UCLA). So, his team varied the technique: Instead of firing their electron beam from one direction at a static crystal, they rotated the crystal and tracked how the diffraction pattern changed. Instead of a single image, they got what was more like molecular computerized tomography scan. That enabled them to get structures from crystals one-billionth the size of those needed for x-ray crystallography.

Gonen says because his interest was in proteins, he never thought much about trying his technique on anything else. But earlier this year, Gonen moved from the Howard Hughes Medical Institute’s Janelia Research Campus in Ashburn, Virginia, to UCLA. There, he teamed up with colleagues, along with Stoltz at Caltech, who wanted to see whether the same approach would work not just with proteins, but with smaller organic molecules. The short answer is it did. On the chemistry preprint server ChemRxiv, the California team reported on Wednesday that when they tried the approach with numerous samples, it worked nearly every time, delivering a resolution on par with x-ray crystallography. The team could even get structures from mixtures of compounds and from materials that had never formally been crystallized and were just scraped off a chemistry purification column. These results all came after just a few minutes of sample preparation and data collection. What’s more, a collaboration of German and Swiss groups independently published similar results using essentially the same technique this week.

“I’ve had dreams in my life where I’m looking through a microscope and I see a molecular model with balls and sticks,” Bertozzi says. “They basically find some microcrystalline schmutz on an EM [sample holder], take some data, and there are the balls and sticks I dreamed about. It’s unbelievable it works so well.”

Because it does work so smoothly, the new technique could revolutionize fields both inside and outside of research, Bertozzi and others say. Tim Grüne, an electron diffraction expert at the Paul Scherrer Institute in Villigen, Switzerland, who led the European group, notes that pharmaceutical companies build massive collections of crystalline compounds, in which they hunt for potential new drugs. But only about one-quarter to one-third of the compounds form crystals big enough for x-ray crystallography. “This will remove a bottleneck and lead to an explosion of structures,” Grüne says. That could speed the search for promising drug leads in tiny samples of exotic plants and fungi. For crime labs, it could help them quickly identify the latest heroin derivatives hitting the streets. And it could even help Olympics officials clean up sports by making it easier to spot vanishingly small amounts of performance-enhancing drugs. All because structures rule—and are now easier than ever to decipher.