Getting proteins to form crystals is only one step for the structural biologist. The next step to sleuthing out a protein's structure involves placing the crystals in an intense beam of X-rays. This radiation bears little resemblance to the broad, diffuse X-rays used in medicine: the powerful X-rays that work best for protein crystallography are produced at giant facilities called synchrotrons, of which only a few dozen exist.
At the Advanced Photon Source synchrotron at Argonne National Laboratory, Illinois, for example, electrons race around a 1.1-kilometre track at close to the speed of light. Radiation generated from the electrons is collected into a 70-metre beamline, which focuses X-rays into a 25-square-micrometre area where crystals can be positioned for analysis.
Proteins in the crystal scatter the X-rays as they pass through, and researchers can decipher a protein's structure from the resulting diffraction pattern. To generate a complete pattern, the crystal must be rotated within the beam so that X-rays pass through in different orientations. The process requires precision: researchers have to collect enough data to solve a structure, while limiting radiation damage to the crystal.
Technologies for manipulating crystals and keeping them at temperatures below 0 °C to decrease radiation damage have got better, but the most dramatic improvement is that experiments can now be done using very small crystals or crystals with many poorly diffracting regions, says So Iwata, who heads the Human Receptor Crystallography Project at the Japan Science and Technology Agency in Kyoto. “Crystals that would have been turned away ten years ago are welcome now,” he says.
Still, a crystal must be as wide as or wider than the beam passing through it to generate a reliable diffraction pattern. That's a problem, because crystals of membrane proteins tend to be small, says Robert Fischetti, a senior scientist at Argonne National Laboratory, which has produced data for crystal structures of several membrane proteins, including the β2-adrenergic receptor (S. G. Rasmussen et al. Nature 450, 383–387; 2007). Technologies such as lipidic cubic phase crystallization have helped researchers to grow crystals, he says, but these are often only 5–10 micrometres across, a tenth the size of most crystals submitted for analysis and much smaller than the X-ray beam used in crystallography studies.
Researchers led by Fischetti have developed a new version of a collimator, a device that blocks most of the X-rays to produce a 'minibeam' of 5 micrometres or less. Collimators, essentially engineered strips of platinum, are placed about 3 centimetres from the sample and are much more than simple pinhole apertures, says Fischetti. “We started out with a single collimator with three parts — the beam-defining pinhole aperture, the capsule around the pinhole and a forward scatter guard tube.”
The first versions of the device caused X-rays to scatter in a way that interfered with the diffraction pattern, but his team has since engineered features, such as a layer of molybdenum, to overcome these problems. They also created double and triple collimators to let users pick the beam size. After one user damaged a collimator by spilling liquid nitrogen on it, they modified the design, eventually making a more robust version with four aperture settings but fewer parts.
What has made the collimator most practical, says Fischetti, is automating the process of selecting the aperture to match the crystal. In the past, technicians had to refocus the beam manually to shrink its size, which took hours. Now, he says, “we can do it in seconds with just clicking”.
Besides allowing the study of smaller crystals, smaller beams let researchers identify the parts of the crystal that diffract better. “In the past, if you had a large crystal that was not homogeneous, you'd look at the crystal and say it was bad,” says Fischetti. “Now, people can find the region that is the best to look at.”