Erica Ollmann Saphire staked her young lab's future on solving the crystal structure of the glycoprotein on the surface of Ebola virus. The move was risky, because the protein is notoriously difficult to express and crystallize in a stable and ordered form. And at least three other groups were racing her for the same solution. Although the high-stakes competition made her “really nervous”, Ollmann Saphire, an immunologist at the Scripps Research Institute in La Jolla, California, and her team won the race and provide a picture of how the deadly virus evades immunity and enters cells.

The work was long and difficult. Over four years, the group expressed 130 versions of the Ebola virus glycoprotein, grew 50,000 crystals, made 30 trips to two synchrotrons to bounce X-rays off the crystal, and tested the diffraction of more than 800 crystals.

What the team found may explain how the virus can lurk for years and still prove so virulent. The structure reveals a 'candy-floss' coating of disordered protein and carbohydrate. The coating keeps the virus stable in the environment and hides the glycoprotein's binding mechanism from attacking host immune cells.

The findings suggest that the glycoprotein acts like a spearfishing line: three attachment subunits are tied together by three fusion subunits, which are wound around them like the line on a spool. Once a possible host contacts the virus, a protease snips the cloak away. Then the attachment subunits are released, triggering the fusion subunits to spear the human cell, allowing infection. “This virus is a fascinating little beast,” says Ollmann Saphire. “It's a stripped-down, tiny machine of just seven genes, with all sorts of structural tricks to replicate itself while evading the immune system.”

The team also managed to solve the structure of an infection-blocking antibody bound to the viral protein. This revealed a vulnerability in the virus — an exposed site on the glycoprotein. The antibody evidently bridged two glycoprotein subunits, somehow preventing it from 'spearing' a human cell. Getting a picture of this site, and a few other sites where antibodies might bind to the glycoprotein, could provide a route to a therapeutic defence against the virus.

Getting that picture proved a “major slog”, says Ollmann Saphire. Of the 50,000 crystals the team grew of the glycoprotein–antibody complex, only one diffracted the X-rays to high enough resolution to close in on the structure. And after identifying this there were still obstacles to overcome, because phase information is not collected during X-ray diffraction work, but is required to assemble the data into a three-dimensional model.

The standard structural biology solution to this problem is either to use a similar, already solved structure or to add a heavy metal to your protein. Neither approach worked here. Instead, with the help of lab tech Ann Hessell, the researchers tried replacing the methionine amino acids in the antibody with selenium-containing methionines. Surprisingly, they got higher-resolution data (see page 177). “We don't know why the selenomethionine antibody worked,” Ollmann Saphire says. “It just did.

Despite worries about putting all her eggs into one difficult-to-decipher basket, Ollmann Saphire says she was driven more by scientific curiosity than by career progression. “We didn't do this to get some big, shiny prize,” she says. “This virus is just truly fascinating, and the structure was a problem that needed to be solved.”