The three-dimensional structures of membrane proteins are notoriously difficult to determine. Raimund Dutzler, a biochemist at the University of Zurich, Switzerland, and his graduate student Ricarda Hilf know this only too well: it has taken them two-and-a half years to resolve the structure of a cation-carrying ion channel from the membrane of the bacterium Erwinia chrysanthemi. This channel, called ELIC, belongs to a large family of ion channels that also includes neuronal ion channels in animals, and is the first of this type for which such a high-resolution crystal structure has been determined.

The first crystal structure of any ion channel — a potassium ion channel — was produced in 1998 by Roderick MacKinnon, a feat for which he received the 2003 Nobel Prize in Chemistry. Dutzler, who was a postdoc in MacKinnon's lab at the Rockefeller Institute, New York, set his sights on a different type of ion channel that opens only when bound by a particular ligand. These 'ligand-gated ion channels' are made up of five protein subunits and include key players in chemical signalling at neural synapses. The best-known members of the family are the nicotinic acetylcholine receptor, which controls muscle movement, and the γ-aminobutyric acid (GABA) receptor, which is involved in learning and memory.

Dutzler chose the bacterial channel as the simplest and smallest example of this type of channel. It is composed of five identical protein subunits, and a bacterial channel protein should also be easier to produce in large amounts. Even so, the duo encountered many setbacks, but Dutzler credits Hilf with being “incredibly persistent and efficient”.

At first they could not produce enough protein to make crystals. “This is usually the end of any crystallography project,” says Dutzler. Undaunted, they fused the channel protein to another protein, called maltose-binding protein, to improve production, and obtained crystals. But these first crystals were not of good enough quality to diffract X-rays at the resolution needed. Dutzler and Hilf had to screen many different crystallization conditions before they were able to make crystals that diffracted X-rays at a resolution of 3.3 ångstroms, the resolution required to identify the positions of individual atoms.

And they weren't done yet. To resolve the atomic structure of a protein, one needs several sets of extremely precise X-ray-diffraction measurements in order to construct the electron-density map that is ultimately used to identify atomic positions.

Fortunately, the team had access to a new and highly sensitive X-ray detector called Pilatus at the Swiss Light Source, the synchrotron facility at the Paul Scherrer Institute in Villigen, where the X-ray data were collected. “We had the most optimal infrastructure. Without it, this project could have taken twice as long,” says Dutzler.

Dutzler and Hilf believe that the structure that finally emerged (see page 375) is conserved among all pentameric ligand-gated ion channels. “It's the first piece of the puzzle,” says Dutzler, adding that they would like to discover what the ligand for this channel is so that they can figure out the mechanics of how it opens to conduct ions.