Structures of a nuclear pore protein, presented by Ivo Melčák, André Hoelz, and Günter Blobel (Rockefeller University, New York, NY), suggest that the pore's central channel expands by an unusual sliding between hydrophilic residues.
Figure 1.
Nup58/45 conformations suggest that its tetramerization interface can slide apart (black arrows). Only one of the two N-helical pairs that generate this interface is shown.
BLOBEL/AAAS
The crystals reveal structures of one of the four nucleoporins that make up the main channel. The authors suggest that the pore is encircled by eight side-by-side tetramers of this nucleoporin, called Nup58/45. The tetramer came in two forms; in one, the dimer–dimer interface was laterally displaced by ∼6 Å compared with the other.
Most protein interfaces depend on hydrophobic residues. But in Nup58/45, an electrostatic dimer–dimer interface permits expansion by allowing alternative hydrogen bond pairings. Hoelz describes this interaction as “the opposite of a leucine zipper.” Intermediate steps that resemble sliding-like movements are probable.
Because each structure was equally abundant in the crystals, the authors propose that little energy is required to switch between the states. Perhaps only the binding of the cargo complex is needed.
If each tetramer is at full extension, the pore diameter might widen by 30 Å, probably during the export of large cargo such as preribosomal subunits. Perhaps the other core pore proteins have similar sliding mechanisms. How accessory proteins alter the situation will be studied for years to come. “There are so many proteins involved,” says Hoelz. “I think that the nuclear pore complex will be full of surprises like [this one].” 
Reference:
Melčák, I., et al. 2007. Science. 315:1729–1732.